Professional Documents
Culture Documents
PHILOSOPHY OF MIND
Series Editor
David J. Chalmers, Australian National University and New York
University
1
3
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide.
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
98765432 1
Printed in the United States of America
on acid-free paper
CONTENTS
Contributors vii
15. Which Causes of an Experience Are Also Objects of the Experience? 351
TOMASZ BUDEK AND KATALIN FARKAS
Index 371
CONTRIBUTORS
Introduction
DOES PERCEPTION HAVE CONTENT?
Berit Brogaard
The title of this volume takes the form of a question, a question which exerts a con-
siderable hold on contemporary philosophers of mind, particularly those working
on perception. This seems to be a fairly new trend. Not so long ago the question
would not even have been considered. Perhaps it would not have seemed intelli-
gible. But things have changed, and there is now a considerable number of articles,
theses, and books aimed at answering it, positively or negatively. What are the fac-
tors responsible for this topic becoming a “live” one? Why is it only now receiving
so much attention?1
I believe the answer to this question is largely historical. On the face of it, tra-
ditional debates about perception were typically concerned with a different ques-
tion, viz., that of whether we perceive the external world directly or indirectly. In
Perception: A Representative Theory, for example, Frank Jackson argues that when
we see things in the environment, we see them in virtue of perceiving something
else. The things that we perceive without having to perceive something else are
sense data. Jackson thought that sense data are something we literally perceive and
the only things we are directly perceptually aware of.
Though the debate about whether we perceive the external world directly or
by virtue of perceiving something else is orthogonal to the debate about whether
perceptual experience has content, it may be argued that the two debates concern
some of the same issues. As we will see below, particular ways of understanding
perceptual content may, at least at first glance, appear to imply that if perception
has content, then the content is an intermediary between the perceiver and the ex-
ternal world, and the perceiver experiences the world by being acquainted with the
content. Things are not quite as simple as this, of course. But it does raise the fol-
lowing question: If the debates about the directness of perception and perceptual
1
Thanks to the anonymous reader for encouraging me to consider these issues and questions. 1
2 Introduction
content are intermingled, what caused the relatively sudden interest in whether
perceptual experience has content?
The notion of perceptual content is not new, of course. In Perception:
A Representative Theory, for example, Jackson casually refers to perceptual con-
tent (e.g., 1977, p. 40), but his endeavors are not aimed at answering the question
of whether perception has content. I believe the recent considerable interest in
the question of whether perception has content may have been a result of the
rise of cognitive science and its focus on the idea of a representational state of
the mind. It seems that the debates in cognitive science have sparked analogous
debates in philosophy of perception about what it means to say that percep-
tual experience has content and whether perceptual experience has content in
the first place. The chapters in this volume deal with these two issues. The first
half of the volume directly addresses the question of whether perception has
content, whereas the second half addresses the question of what the content of
perceptual experience is like.
The question of whether perceptual experience has content may, at first
glance, seem rather trivial. A simple argument for the view would run as follows.
Perceptual experience is accurate or inaccurate. If it’s accurate, it’s accurate in
virtue of some proposition p being true. If it’s inaccurate, it’s inaccurate in virtue of
some proposition p being false. But that proposition p just is the content of percep-
tual experience. So perceptual experience has content. While this argument has
something to be said for it, it doesn’t quite get to the core of the debate.
As Susanna Siegel points out in her book The Contents of Visual Experience,
one flaw in the argument from accuracy conditions is that it does not require that
the accuracy conditions had by experiences are conveyed to the subject by her
experience (Siegel 2010, p. 43). But accuracy conditions that are not conveyed to
the subject by the experience are not suitable to serve as experiential contents. For
example, all experiences are accurate if and only if Mother Nature corroborates
the experience. But no typical experiences convey Mother Nature corroborates my
experience to the subject and the proposition Mother Nature corroborates my expe-
rience ought not normally count as the experiential content. So, the general move
from accuracy conditions to contents is invalid.
Siegel (2010) distinguishes between weak and strong perceptual content. On
the weakest acceptable formulation of the view that perceptual experience has
content, experiential content is its accuracy conditions properly conveyed to the
subject.
Weak Content View
Experience e has the proposition p as a content iff p is conveyed to the subject
of e, and necessarily if e is accurate, then p is true.
It is the question of whether perceptual experience has content in this mini-
mally acceptable sense that directly or indirectly has inspired the topics of most of
the chapters in this volume.
Introduction 3
As we will see below, it takes substantial argument to establish this view with
some degree of plausibility, and many thinkers reject the view, including naïve
realists and enactivists.
There are also historical views not discussed in this volume that, at least prima
facie, are at odds with the weak content view. On one such view, perceptual ex-
perience is a strictly non-intentional (or non-representational) kind of sensation
that requires further interpretation in order to be fully graspable by the perceiver.
We might call this type of view the ‘raw sensations view’. Thomas Reid’s two-state
view can be interpreted along these lines. Because raw sensations are strictly
non-intentional, they do not have accuracy conditions at all. So unless they are
subjected to further interpretation, they cannot properly be said to have content
even in a weak sense.
Visual form agnosia, a brain condition in which perceivers can see that there
is something in front of them but cannot identify what is in front of them, may
shed some light on what raw sensations are. Visual agnosia patients sometimes de-
scribe the “something” in front of them as a blob without clear boundaries, color,
shape, or texture. It’s not implausible to think that only raw sensations are avail-
able to these patients’ conscious visual system. The information they consciously
possess about their environment does not represent any particular thing but just
a “something.” Arguably, the accurate/inaccurate dichotomy does not apply to the
visual experience of visual agnosia patients. Their visual experiences certainly are
not like normal, accurate perceptual experiences but nor is it the perceptions (in
the standard sense) that have gone wrong.
Though the weak content view is already very contentious, some thinkers be-
lieve that being contentful is essential to perceptual experience. On this view, it’s
a fundamental feature of perceptual experience that it has representational con-
tent, i.e., content that is suitable to serve as the content of a propositional attitude
(Siegel 2010).
Strong Content View
Experience e has the proposition p as a content iff necessarily, the subject of e
bears a propositional attitude towards p.
The strong content view, if true, appears to rule out a number of familiar views
about perceptual experience. On the face of it, views that take perception to be
a perceptual relation to an external object (and features of that object) appear to
be at odds with the strong content view. Naïve realists often hold that perception
is fundamentally characterized by its relational properties but not fundamentally
characterized by its representational properties (see, e.g., Fish 2009). Veridical (or
non-hallucinatory) perception is a perceptual relation to an object that does not
involve representational properties in any interesting sense. On this view, verid-
ical perception has content only in the weak sense: when a perceiver stands in the
perceptual relation, she is in a position to acquire knowledge on the basis thereof.
Disjunctivism adds to this view that cases of hallucinations (Langsam 1997;
4 Introduction
Snowdon 1980/81; Smith) and in some cases also illusions (Hinton 1973; McDowell
1982, 1986, 2008; Martin 2002; Fish 2008, 2009) are not of the same fundamental
kind as veridical perceptions.2 Some hold that hallucinations are thoughts or
beliefs (e.g., Fish), others that they are more closely related to mental imagery
in that they are relations to imaginary objects (e.g., Smith 2002), and yet others
that nothing more can be said about hallucinations than that they are internally
indiscriminable from veridical perception but that the perceptual relation fails to
obtain (Martin). Naïve realists can thus hold that hallucinations, and perhaps also
illusions, have representational content and yet deny that it is a fundamental fea-
ture of perceptual experiences that they have representational contents.
Though naïve realists with traditionalist leanings adamantly deny that percep-
tual experiences have any type of content, some thinkers are more all-embracing.
As we will see, several of the contributors to this volume argue that naïve realism,
initial appearances to the contrary, can accept the strong content view.
There are also historical views that appear to be at odds with the strong content
view. One is adverbialism. The name ‘adverbialism’ derives from the word group
made up of adverbs. Adverbs modify the verb that they grammatically are adja-
cent to. They may lead to further specification of features of the denotation of the
verb or an alteration of the meaning of what was said. For example, if I am told
that John made a sandwich, I might be interested in knowing whether he made it
slowly, gracefully, angrily, or carelessly. ‘Slowly’, ‘gracefully’, ‘angrily’, and ‘carelessly’
are adverbs that provide information about features of the action picked out by the
verb. Adding these adverbs to the original sentence does not cancel out the original
meaning. If John made a sandwich angrily, then he made a sandwich. However,
other adverbs will alter the meaning of the original sentence when added. For ex-
ample, the sentence ‘John barely made a sandwich’ doesn’t entail that John made a
sandwich. Adverbialism holds that features of perception play a role analogous to
the first group of adverbs (Chisholm 1957). They specify a way in which the subject
perceives the world. For example, if John has a perceptual experience of a red cat,
then John is appeared to redly and cat-wise. On this view, perception is neither a
relation nor a mental state with content in the strong sense. It just is a way of per-
ceiving. Roderick Chisholm (1957) is a legendary defender of the adverbial theory.
Wylie Breckenridge (n.d.) is one of its contemporary defenders.
The sense-datum view, originally defended by Russell (1912), Broad (1925),
Price (1950), Ayer (1956), and Jackson (1977), also appears compelled to deny that
perception has content in the strong sense.3 On the sense-datum view, perceivers
don’t perceive the world directly; instead they perceive sense-data by standing in
a perceptual relation to these sense-data. Sense-data are proxies for objects in the
external world. They have colors, shapes, textures, and so on. While sense-data
2
Fish holds the intermediate view that illusions are sometimes good cases of perception and
sometimes bad. See Fish (2009) and Brogaard (2011).
3
More recent defenders include: Robinson (1994), and Casullo (1987).
Introduction 5
represent objects and features in the external world, sense-data are not them-
selves perceptual experiences. Perceptual experiences fundamentally consist in
being related to sense-data.
The aforementioned views are among the best known views that commonly
are thought to deny that perceptual experience has content in a strong sense.
What of views that embrace the strong content view? One candidate to be a view
of this kind is representationalism (or intentionalism, as it’s sometimes called).
Strong representationalism is the view that for a perceptual experience to have
a certain content just is for it to have a certain phenomenal character (Chalmers
2004). A weaker version merely holds that the contents of perceptual experience
supervene on the phenomenal character of the experience. Though representa-
tionalism makes more commitments than the strong content view, the two are
compatible. Representationalism, as well as the strong content view, holds that
perceptual experience fundamentally consists in representing externally instanti-
ated properties and/or objects.
In the following I will look in some detail at the contributions to this volume
and, along the way, point to some potential issues raised by each contribution.
FIGURE 1.1 The Ebbinghaus illusion Studies have shown that this illusion leads to a
misperception of the size of the central circle but only marginally affects grasping behavior
directed at the central circle.
The dorsal stream, which starts in the visual cortex and then runs upwards into
the parietal cortex and ends in the sensorimotor cortex, is in charge of guiding
ongoing action.
In neurotypical humans, the ventral and dorsal visual systems need not
ascribe the same properties to objects. One example of a case in which the two
pathways ascribe different properties is the Ebbinghaus illusion (figure 1.1), in
which a circle that is surrounded by smaller circles looks larger than a circle of
the same size that is surrounded by larger circles. There is empirical evidence
that the two visual streams attribute different size properties to the inner circle.
While the circle that is surrounded by the smaller circles looks larger than a
same-sized circle surrounded by larger circles, our grip-size is not influenced
by the illusion. Our hand aperture when attempting to reach to and grasp the
middle circle is the same in the two cases.
The representational view, Nanay argues, has a straightforward way of
explaining the different property attributions. The two visual pathways represent
the inner circle as having two different size properties.
According to Nanay, the anti-representationalist cannot offer the same expla-
nation. If perception is a perceptual relation between a perceiver and perceptible
property instances, then there is a single perceptual relation between the perceiver
and the property instances regardless of how the brain processes the relevant in-
formation. In a three-dimensional model of the Ebbinghaus illusion, the percep-
tual relation is a relation between the perceiver and the property instances of a
poker chip. According to the enactivist, the information that is relevant in percep-
tion is stored in the external world. But, Nanay argues, it is unclear which of the
properties is relevant to perception in the case of the Ebbinghaus illusion.
It would thus seem that the anti-representationalist must deny that there are
two perceptual episodes in the Ebbinghaus scenario. However, denying this has dire
consequences. The size property the middle poker chip appears to have plays a role
in the justification of our beliefs, and the property that our hand aperture is tracking
plays a role in our ongoing actions. So, neither can be dismissed as irrelevant.
Introduction 7
of that experience is a sense that when one’s situation deviates from some
optimal body-environment relationship, one’s activity takes one closer to that
optimum and thereby relieves the ‘tension’ of the deviation. One does not
need to know what that optimum is. One’s body is simply solicited by the situ-
ation to get into equilibrium with it. (Dreyfus 2002, p. 12)
Siegel considers several challenges to the weak content view: either experienced
mandates do not involve any content at all, or experienced mandates do involve
content but the content does not play any explanatory role in action.
Siegel argues that experienced mandates have contents, and claims to identify
the content that could reflect the distinctive character of experienced mandates.
She offers the weak content view as a premise in one of her arguments against the
idea that perceptual content plays no significant role in explaining behavior. A
simplified version of Siegel’s argument for the weak content view can be articu-
lated as follows (see Siegel 2010 for the full argument):
Siegel’s Argument for Weak Content
1. All visual (perceptual) experiences present clusters of properties F as
being instantiated.
2. So, necessarily, things are as E presents them to be only if F is
instantiated. (from 1)
3. So, E has a set of accuracy conditions C, conveyed to the subject of E.
(from 2)
Conclusion: All visual experiences have content.
Below we will consider some of Heather Logue’s concerns about the argument. As
I see it, one potential problem with the argument is that it doesn’t make it explicit
what ‘present’ and ‘convey’ mean and how the two are connected. As I understand
Siegel, a perceptual experience presents a property as instantiated only if it percep-
tually seems that way to the subject. Likewise, F’s being instantiated is conveyed to
a subject only if things perceptually seem that way to her. So, the argument could
be formulated as follows:
The Argument from Appearances
1. All visual (perceptual) experiences make it perceptually seem to the
subject that clusters of properties F are instantiated.
2. So, necessarily, things are as they perceptually seem only if F is
instantiated. (from 1)
3. So, E has a set of accuracy conditions C that correspond to how things
perceptually seem to the subject. (from 2)
Conclusion: All visual (perceptual) experiences have content.
One potential issue with this version of the argument lies with the step from
premise 2 to premise 3. The accuracy conditions for an experience should be the
complete condition C such that the experience is accurate if and only if C obtains.
However, we don’t get that from premise 2. Premise 2 only mentions properties
Introduction 9
that perceptually appear to be instantiated. If, however, one doesn’t pay much at-
tention to a tiny leaf on a tree in one’s visual field, then it may not perceptually
seem to one that properties of that leaf are instantiated. Whether or not the leaf ’s
properties are instantiated, however, might impact the accuracy of the experience.
So, we may not be able to get the accuracy conditions, or the experimental con-
tent, from how things perceptually appear to the subject. Of course, if perceptual
appearances just are perceptual experiences, then this objection has no traction.
Kathrin Glüer: In “Looks, Reasons, and Experiences,” Kathrin Glüer starts off with
an argument for the strong content view. According to Glüer, contents either are, or
are essentially such that they determine, accuracy conditions. Accuracy conditions, she
says, are conditions the world must satisfy in order for the content in question to be true.
Contents have their truth-conditions essentially. They play a role in accounts of how
different mental states relate to each other. We assign contents to psychological states
to model a specific kind of structure among them and to predict and explain how one
state can lead to another state. According to Glüer, the overarching reason perceptual
experiences have contents of this sort is that they are propositional attitudes. Perceptual
experiences, she holds, are beliefs. Since beliefs uncontroversially have contents, and
perceptual experiences are beliefs, perceptual experiences have contents.
To say that perceptual experiences are beliefs is not to say that there are no
differences between perceptual experiences and other beliefs. For example, per-
ceptual experiences have a phenomenology that is notably different from the phe-
nomenology of beliefs that are not perceptual experiences. Perceptual experiences
have a distinctive sensory phenomenology.
One challenge for the view that perceptual experience is belief is to explain
cases in which we don’t believe what we experience. When we dip a stick in water
and it looks bent, we don’t come to believe that it is bent. We know that it’s not.
Alex Byrne (2009) has argued that perceptual experiences are primitive beliefs.
According to him, we do indeed believe that the stick is bent on a very primitive
level. But on a more rational level, we do not believe that the stick is bent.
This is not Glüer’s strategy. According to Glüer, when we have a perceptual ex-
perience, things look a certain way to us. Gluer takes these looks to constitute the
content of perceptual experience. If I am looking at a blue car, and the car looks blue
to me, then the content of my perceptual experience is ‘Look(the car is blue)’, where
‘look’ is an operator on the embedded material. When I look at the stick in the
water, I come to believe that it looks bent but I don’t come to believe that it is bent.
Glüer considers and replies to a potential problem for her appearance account
of perceptual experiences. The envisaged counter-argument runs as follows. The
phenomenal notion of look cannot be used to specify the very content of visual ex-
perience because ‘look’ is a propositional attitude operator and thus cannot occur
in the content of any first-order propositional attitude. Glüer calls this the ‘Attitude
Operator Argument’. The argument can be summarized as follows:
The Attitude Operator Argument
1. Perceptual experience is a first-order propositional attitude.
Introduction
10
The first premise is widely held to be true: second-order propositional attitudes are
not typically considered perceptual states. For example, most views of perception
would hold that you cannot perceptually experience having a belief or a desire.
That process would count as introspection, not perception. Of course, those who
take introspection to be a form of perception would resist drawing this distinction.
But even people in this camp can admit that the standard forms of perception that
Glüer is interested in are first-order propositional attitudes and not second order.
Glüer rejects premise 2. ‘Look’, she argues, is not a propositional attitude oper-
ator. So, perceptual contents can be first-order, even if they contain a ‘look’ operator.
The reason, she argues, is simple: Propositional attitude operators create hyperinten-
sional contexts. But ‘look’ does not. So, ‘look’ is not a propositional attitude operator:
Glüer’s Counterargument
1. If Φ is a propositional attitude operator, then it creates a hyperintensional
context.
2. ‘Look’ does not create a hyperintensional context.
Conclusion: ‘Look’ is not a propositional attitude operator.
‘Superman’ and ‘Clark Kent’ are necessarily co-referential but substituting one for
the other appears to elicit a change in truth-value. This indicates that ‘look’ gener-
ates a hyperintensional context. However, Glüer thinks this appearance is illusory.
According to Glüer, ‘look’ satisfies the following substitution principle:
Substitution Principle
Co-phenomenal expressions can be substituted salva veritate in ‘look’-contexts.
I take it that Glüer takes co-phenomenal expressions to be expressions that refer
to entities that look the same to a perceiver in normal viewing conditions. The
Substitution Principle implies that if two expressions are not co-phenomenal,
then substituting one for the other will change the truth-value of sentence.
Glüer notes that a Superman-look is very different from a Clark Kent-look.
So, ‘Superman’ and ‘Clark Kent’, though co-referential, are not co-phenomenal.
Introduction 11
That seems right. But Glüer then argues that it follows from this that ‘look’-
contexts are not hyperintensional. This, however, could be questioned. On the
standard definition of ‘hyperintensionality’, an operator is hyperintensional just
in case substituting an expression for a logically (or metaphysically) equivalent
expression under the operator can change the truth-value of the whole. On that
definition, ‘look’ generates hyperintensional contexts.
However, there might be an independent problem with the second premise of
the Attitude Operator Argument:
Premise 2
If perceptual contents contain ‘look’ operators, then perceptual experience is
not a first-order propositional attitude.
This premise appears to rest on the assumption that hyperintensional operators
are propositional attitude operators. This latter assumption, however, is mistaken.
Fictional operators, such as ‘according to the Sherlock Holmes stories’ are hyper-
intensional yet they are not propositional attitude operators. So, if premise 2 is
true, then it is not because ‘look’ is hyperintensional. For the Attitude Operator
Argument to be effective against Glüer’s position, then, one would need to show
on independent grounds that ‘look’ operators are propositional attitude operators.
Mark Johnston: In his contribution to this volume, Mark Johnston argues against the
view that perceptual experience has content. One problem with the view, Johnston
argues, is that it fails to adequately capture the distinction between illusory and
non-illusory experience. As there can be veridical illusions, the distinction be-
tween veridical/non-veridical experience is orthogonal to the distinction between
illusory/non-illusory experience. Johnston considers a number of optical illusions
that bear on the latter distinction. In the larger ball illusion, for example, two balls
are presented as located behind each other in a three-dimensional room. Because
the two balls actually have the same size, the ball that is presented as being further
away looks much larger than the other ball. Johnston points out that there is no
non-illusory counterpart of the illusion. If we could somehow override the standard
perceptual mechanisms that present relative sizes, we might experience the two balls
as having the same size. But it would still be an illusion because the perceptual sys-
tem then wouldn’t function normally. But the content view fails to account for this.
If we were to ask “why is the experience illusory?” the answer could not be that the
content misrepresents, because it doesn’t. The explanation would have to appeal to
the failed causal relation between the object and the perceiver. But on the content
view, the causal relation is not a constituent of the content of the experience, which is
to say, the content plays no role in the explanation. So, even if the content view is able
Introduction
12
Even if we assume that the environment is hospitable (e.g., there is no fake lighting
or deceitful plastic surgery), Travis would probably balk at the move from 2 to
3. This is because seemings that determine a unique content rely on the agent’s
rational abilities. But seemings that depend on the agent’s rational abilities cannot
serve as the source of content for visual experiences.
One might question Travis’ use of appear words. An agent’s cognitive system
may determine that a Porsche painted chrome yellow but covered in mud ought to
be said to look yellow. In these circumstances it would be true to reply to someone
who asks ‘What does Pia’s Porsche look like?’ with ‘Yellow’. This, however, is what
Roderick Chisholm (1957) calls the comparative use of ‘look’ (see also Jackson 1977).
‘Pia’s Porsche looks yellow in daylight’ cashes out to ‘Pia’s Porsche looks one of the
ways yellow Porsches look in daylight’. Following standard linguistic analysis, this
can be analyzed as ‘there is an x such that x is a way yellow Porches look, and Pia’s
Porsche looks x’ (Brogaard 2012a, b). This analysis makes unreduced appeal to a no-
tion of ‘look x’. This latter notion of ‘look’ is the non-comparative (perceptual) use.
In the circumstances in which Pia’s Porsche is covered in mud, it non-comparatively
looks beige. So, one might argue, the non-comparative looks do determine a par-
ticular set of accuracy conditions, independently of the agent’s rational abilities.
However, this reply does not get to the heart of the matter. Consider a per-
ceptual experience of a white wall that is partly illuminated by sunlight. On a very
restricted notion of non-comparative looks, the wall does not look uniformly col-
ored. However, thinkers who hold that the experiences represent would want to
say that the experience of the wall represents the wall as white. More generally, if
perceptual experiences represent, they ought to represent color constancies as well
as size and shape constancies. But it follows, then, that perceptual seemings cannot
be associated with one set of accuracy conditions independently of how our brains
happen to calculate these color-, size-, and shape-constancies.
Diana Raffman: In “Disjunctivism, Discrimination, and Categorization”,
Diana Raffman starts out by exposing some misconceptions of the perceptual
indiscriminability relation as it figures in recent treatments of disjunctivism by
Martin (2004, 2006) and Siegel (2004). As Siegel points out, one needs a way to
characterize the relevant pairs of perceptions and hallucinations to state the debate
between disjunctivism and its opponents. Introducing the notion of indiscrim-
inability has been the standard way of meeting this need. Raffman, however, is
skeptical that this relation can satisfy the need.
Introduction 15
pairs of patches have the same color, they nonetheless adjust the color measure as
they proceed along the continuum. Raffman suggested that this indicates that the
participants fail to notice the changes despite perceiving them. However, if the
changes were consciously perceived, then this proposal seems to rest on a con-
troversial assumption to the effect that one can have perceptual experience in the
absence of attention.
Even if perceptual experience can overflow selective attention, Raffman’s in-
terpretation could be questioned. The subjects in the study report that they don’t
perceive a change in color but nonetheless adjust the color measure as they move
about in the perceptual continuum. Their adjustment of the color measure could
reflect a sensitivity to wavelength in the visual dorsal stream. Milner and Goodale
originally took dorsal stream processes to be responsible for blindsight—the ability
had by people with V1 lesions to reliably report color and other attributes of visual
stimuli that are not consciously seen (Goodale et al. 1991; Goodale and Milner 1992).
This hypothesis has later been countered. However, the question of the extent to
which the dorsal pathway can discriminate among color stimuli remains largely
unexplored.
Relational View
Perceptual experience is fundamentally a matter of being perceptually related
to objects in the environment.
The two views are commonly thought to be in opposition, because it is assumed that
if perception is fundamentally characterized by its representational properties, then it
cannot be fundamentally characterized by its relational properties. It is this common
belief that Schellenberg takes issue with in her chapter. While Schellenberg defends
both views, others reject the whole package. Sense-data theorists, for example, take
perception to be a relation between a perceiver and sense-data, and adverbialists
take perception to be an act of perceiving in a certain way. So, sense-data theorists
and adverbialists would have to reject both views as stated.
Introduction 17
Schellenberg distinguishes the representational view from what she calls the
‘Association Theory’.
Association Theory
Every experience can be associated with (propositional) content in the sense
that sentences can be articulated that describe how the environment seems
to the subject, without the content expressed being a proper part of the
experience.
According to Schellenberg, virtually any theory of perceptual experience could
accept the association theory, as it merely requires that we can use language to
partially describe our perceptual appearances. The association theory is akin to the
weak content view. The main difference is that the association theory specifically
mentions how content comes to have accuracy conditions, viz., through descrip-
tion sentences.
After presenting an argument for the association theory, Schellenberg goes on
to defend the representational view. In broad outline, the argument can be sum-
marized as follows. In order for things to perceptually seem a certain way to us, we
need to employ discriminatory selective capacities that constitute the seeming. For
example, if it seems to me that the fire truck is red, I must be able to discriminate
red from green. But employing these capacities in this way just is to represent the
environment in virtue of using these capacities. So, Schellenberg argues, percep-
tual experience is fundamentally a matter of representing the environment in a
certain way, which is the first principle.
The step from the last premise in this simplified presentation of her argument
to the conclusion rests on a particular epistemic view about perceptual experi-
ence. It seems incontestable that to use certain discriminatory selective capacities
that constitute seemings just is to represent one’s environment in a certain way.
However, one might question that this shows that it’s the perceptual experience as
opposed to the state of seeming itself that represents the environment. To get to the
conclusion that perceptual experience fundamentally is a matter of representing
the environment a certain way, it could be argued that a further premise is needed,
viz., the premise that if certain discriminatory selective capacities are required in
order for a certain state of perceptual seeming to obtain, then these discriminatory
selective capacities are also required in order for the underlying perceptual experi-
ence to obtain.
One might object to this last implicit premise. On the liberal view, perceptual
experience can represent high-level properties such as being a vine leaf maple,
being a cork screw, being sad, and being English. If the liberal view is false, then
I cannot have a perceptual experience that represents a vine leaf maple. But it can
nonetheless still perceptually seem to me that the tree is a vine leaf maple. Certain
discriminatory selective capacities, then, are required in order for a certain state
of perceptual seeming to obtain. But, if the liberal view is false, then it doesn’t
Introduction
18
follow that the same discriminatory selective capacities are required in order for
the underlying perceptual experience to obtain.
A simple move to avoid this consequence would be to adopt a liberal view
with respect to perceptual properties. While some would be happy with this as-
sumption (Siegel 2005), others might think the liberal view is too controversial to
assume up front (see, e.g., Brogaard 2013). Siegel (2005) offers an argument for the
liberal view. However, her argument does not show that it’s perceptual experiences
as opposed to perceptual seemings that represent high-level properties. So it is not
conclusive in the present context. I return briefly to the debate about high-level
properties below when discussing William Lycan’s chapter.
Heather Logue: In her contribution to this volume, Heather Logue defends a
version of the weak content view as well as the claim that three different content
views of varying strengths are compatible with naïve realism. The three content
views can be characterized as subsets of the following four assumptions:
1. There is a proposition associated with E, and
2. This proposition specifies the way things perceptually appear to the
subject in virtue of having E.
3. Perceptual experience consists in the subject perceptually representing her
environment as being a certain way.
4. Perceptual experience fundamentally consists in the subject perceptually
representing her environment as being a certain way.
According to Logue, (1) and (2) are the bare minimum that one is committed to when
one says that experience has content. Assumptions (1)‒(3) constitute the medium con-
tent view and (1)‒(4) constitute the strong content view. Assumption (4)is stronger
than (3) because it presupposes that some element of perceptual experience, for ex-
ample, its phenomenal character or some aspect of its epistemological role supervenes
on the subject perceptually representing her environment as being a certain way.
Logue argues that one of the main reasons naïve realists are reluctant to accept
the view that perceptual experience has content originates in their belief that accept-
ing this position would make naïve realism redundant. They seem to think that the
assumption that perceptual experience has content (even in a minimal sense) is suf-
ficient to explain its epistemological, behavioral, and phenomenal features. Because
naïve realists believe that accepting even the weak content view makes naïve realism
redundant, they are willing to bite the bullet and say that there is no proposition asso-
ciated with perceptual experience that specifies the way things perceptually appear.
It may seem that beliefs grounded in perceptual experience ought to have
contents that specify the way things perceptually appear. If this is so, and all per-
ceptual experiences give rise to (implicit or explicit) beliefs, then we can say that
the contents of beliefs grounded in perceptual experiences compose the contents
of those experiences. This would be a variation on Reid’s (1983) position.
However, as Logue points out, beliefs based on perceptual experience bear
not just on experience but also on background beliefs. So, we cannot read off the
Introduction 19
FIGURE 1.2 The Müller-Lyer illusion In the Müller-Lyer illusion you believe the lines are of the
same lengths but no matter how long you look, your continue to experience them as having
different lengths. This illustrates a case in which perceptual information is encapsulated from
belief influence.
Introduction
20
For example, why does my experience of a blue car in good viewing conditions
give rise to the belief that the car is blue and not to the belief that the car is black?
The most natural answer to this question is that perceptual experience is associ-
ated with a proposition that captures how things perceptually appear to be, and
that this proposition constrains the content of my belief. According to Logue,
naïve realists are more likely to be willing to accept that some perceptual experi-
ences give rise to beliefs than they are to accept that perceptual experiences are
assessable for accuracy.
Logue doesn’t think that accepting the weak content view shows that there
is no room for naïve realism. According to her, the assumption that perceptual
experience has content (even in a minimal sense) is unlikely to be sufficient to
explain its epistemological, functional, and phenomenal features. For a strong rep-
resentationalist, the content of experience is sufficient to explain the phenomenal
features of experience. But another option is to say that its phenomenal features
derive from the subject standing in certain perceptual relations to entities in her
environment. If this is so, then veridical experience fundamentally consists both in
the subject’s perceptually representing her environment as being a certain way and
as being perceptually related to entities in her environment. According to Logue,
this last option amounts to a reconciliation of naïve realism with the strong con-
tent view.
Logue’s argument from belief generation, as formulated, only shows that
the weak content view is correct. However, it might be extended to provide
evidence for the strong content view: If things perceptually appear to the sub-
ject a certain way, we can say that the subject perceptually represents her en-
vironment as being a certain way, perhaps in virtue of having certain selective
discriminatory capacities, as Schellenberg argues. For example, if a blue car
perceptually appears to me to be blue, then I perceptually represent a car as
being blue in virtue of my abilities to discriminate between blue and other
colors. Moreover, if things perceptually appear to the subject a certain way,
then she is more likely to form a particular belief rather than another. For
example, if a blue car perceptually appears to me to be blue, then I am more
likely to form the belief that the car is blue than the belief that the car is some
other color. But an aspect of the epistemological role of experience is to con-
strain the content of belief originating in the experience. So, this aspect of the
epistemological role of experience is grounded in the way the subject percep-
tually represents her environment as being. In other words, if the content of
perceptual experience is what constrains the contents of beliefs originating
in the experience, an aspect of the epistemological role of perceptual experi-
ence supervenes on the subject perceptually representing her environment as
being a certain way. So, as Logue cashes out the meaning of ‘fundamentally’,
perceptual experience fundamentally consists in the subject perceptually rep-
resenting her environment as being a certain way. So, the strong content view
is correct.
Introduction 21
white squares). But this raises the question of why perceptual experiences give
rise to particular perceptual seemings and not others. The most natural answer
to this question is that perceptual experience is associated with a proposition that
constrains how things perceptually appear. Adapting the further argument for the
strong content view requires the further premise that perceptual seemings, un-
like perceptual experiences, are rationally based. But this is a natural stance to
take when maintaining that perceptual seemings and perceptual experiences come
apart (Sosa 2009a, 2009b).
Benj Hellie: In “Love in the Time of Cholera,” Benj Hellie argues for a dif-
ferent type of reconciliatory view that’s naïve-realism friendly in some but not all
respects.4 He takes perceptual states to be ecological states rather than inner, or
psychological, states but holds that all perceptual states have content in a minimal
sense. It is in this latter respect that his view is unfriendly to naïve realism, or at
least to those among naïve realists who deny the content thesis.
Hellie is not a big fan of the phrase ‘perceptual experience’. But if we need
to use it, he suggests that we take it to refer to a certain aspect of the stream
of consciousness, viz., that aspect which is one’s course of attentive stances.
The attentive stance delimits the perceptual part or aspect of the stream of
consciousness.
One of Hellie’s aims in his chapter is to refute a content-based case against
naïve realism. The argument against naïve realism that Hellie takes issue with can
be summarized as follows. Regardless of what naïve realists take hallucinations
to be, most are willing to admit that it can perceptually seem that one is engaged
with the environment. This seeming, however, can be inaccurate. So, the seeming
has content. As both veridical and non-veridical experience have content if one of
them does, naïve realism is false.
Hellie replies to this argument that it mistakenly assumes that perceptual
appearances are an aspect of perceptual experience. One of the arguments he pro-
vides is an argument from linguistics. According to Hellie, the perceptual verbs
‘look’, ‘feel’, ‘taste’, ‘smell’, and ‘sound’ are copular verbs just like ‘be’ and ‘become’.5
These verbs take an adjectival predicate as its their syntactic complements, as in:
(1)
(a) Andrea is/becomes/looks tall.
(b) Kim is/becomes/looks similar to a cat.
(c) Luke is/becomes/looks like a dog.
4
Hellie talks about direct realism, but I use the phrase ‘naïve realism’ here in the interest of a
uniform terminology.
5
Note that subject-raising verbs and copular verbs are sometimes all classified as copular verbs.
I don’t have a problem with this standard classification. If, however, ‘look’ is a copular verb, then it
should be more similar semantically to quintessential copular verbs, such as ‘is’ and ‘become’, than to
subject-raising verbs.
Introduction 23
The perceptual copular verbs, Hellie points out, resist taking ‘that’ clauses as their
complements and only reluctantly take non-finite verb phrases (e.g., ‘to have had
a good time’) as their complements. In this respect they behave differently from
subject-raising verbs, such as ‘seem’, ‘appear’, ‘believe’, ‘prove’, ‘expect’, ‘turn out’,
‘find’, ‘deem’, and ‘assume’.
(2)
(a) It appears/seems that Sam is running for office.
(b) #It looks that Sam is running for office.
(c) My shoes seem to have been left out in the rain.
(d) ?My shoes look to have been left out in the rain.
Hellie takes this to suggest that the perceptual copular verbs do not operate syn-
tactically on clauses but on predicates, which means that they do not operate se-
mantically on propositions but on properties. Hellie proceeds from these linguistic
considerations to an account of perceptual copular verbs as expressing special
kinds of beliefs.
In previous work I have argued that ‘seem’, ‘look’, and ‘appear’ all function as
subject-raising verbs (see, e.g., Brogaard 2012b). Though I grant that ‘look’ does
not take ‘that’-clauses as its complement, I do not think this observation suffices
for treating it as semantically different from ‘seem’ and ‘appear’. I take this be-
havior of ‘look’ to be an irregularity of the verb. Even if Hellie is right that ‘look’
is a true copular verb, however, he does not provide a knockdown argument for
treating perceptual seemings or appearances as beliefs. The main reason for this is
that even though perceptual seemings can be expressed in terms of ‘look’, they are
equally well expressed in terms of ‘seem’ or ‘appear’, which uncontroversially are
subject-raising verbs.
Though the reconciliatory views defended by Schellenberg, Logue, and Hellie
are sympathetic to the content views, they may still face some of the same chal-
lenges as the strong versions of naïve realism. By insisting that perceptual experi-
ence fundamentally is a matter of being perceptually related to objects and features
in the environment, they may face problems along the lines of those raised by
Bence Nanay (this volume). By holding that perceptual experience fundamentally
is a matter of being related to the environment, they need to specify in which ways
ventral and dorsal stream representations differ in this regard. The visual ventral
and dorsal pathways appear to attribute different properties to objects in the envi-
ronment. Does this mean that there are two ways of being related to the environ-
ment: A dorsal way and a ventral way? Or is one type of vision not fundamentally
a matter of being related to the environment?
There are also ways in which the reconciliatory accounts may fare better than
their naïve realist cousins. For example, they can agree with Nanay that standing
in perceptual relations to the environment presupposes integrating multisensory
information in a representational form. Given a reconciliatory view, it is unprob-
lematic to grant that this is so.
Introduction
24
4,810 meters high. If, however, there is no concrete particular to serve as the first
constituent of the pair, what takes its place?
There are two possible answers to this question. Either nothing takes its place
or something takes its place. If nothing takes its place, then it would seem that
the propositions expressed by sentences containing empty names and hallucina-
tions are properties. However, we can quickly rule out this possibility, as properties
and propositions are of different semantic types. If, on the other hand, some sort
of placeholder takes its place, then the proposition cannot be the content of the
mental state or sentence in question. For example, suppose the proposition has
the form of a propositional function Fx. Theorists who hold that sentences express
propositional functions, such as Irene Heim (2006; see also Brogaard 2012a), take
the functions to be existentially closed by default. So, we would get ∃x(Fx). But
this says that something is F. Surely, the content of my hallucination of a red spider
is not the proposition that something is red. If, on the other hand, x is a variable,
a concrete-particular hole or a concrete-particular spatio-temporal location, then
the content says that the variable or hole or spatio-temporal location x is red. This,
too, cannot be the content of my hallucination of a red spider.
Based on these types of considerations Tye dismisses his old view that the
contents of perception are Russellian propositions and proposes two alternatives.
The first is that the contents of perception are sets of possible worlds. This view
is very similar to the view that perception has scenario content (Peacocke, 1983,
1992). Scenario content is normally thought to be a set of centered worlds, spe-
cifically worlds that include a perspective. On this view, the content of a veridical
perception of a particular red spider is the set of possible worlds that contains that
particular red spider. In the case of a hallucination of a red spider, there is no red
particular red spider. So, there is no set of worlds that can constitute the content.
The content, then, is the empty set.
The other suggestion Tye makes is that propositions are structured but that
they are not composed out of properties and/or objects. Instead they are com-
posed out of Kaplanian characters. David Kaplan (1989) proposes that expressions
have a (semantic) character that determines their semantic value. Characters are
functions from a context of utterance to a semantic value. For example, indexicals
such as ‘I’ and ‘now’ are functions from a context of utterance to the speaker and
the time of the context and proper names are functions from a context to the ref-
erent of the proper name.
If we apply this idea to the case of perception, we can say that the content of
a perceptual experience of an object that has a certain property would consist of a
function from the perceptual situation to the object that corresponds to the phe-
nomenal character and a function from the perceptual situation to the property
that corresponds to the phenomenal character. The content of a veridical or hallu-
cinatory perception of a red spider, then, is a structured proposition that consists
of a function from a perceptual situation to the object that corresponds to the
Introduction
26
than three men on the bench’. The objects and their features are presented as spa-
tially and temporally located relative to each other.
One virtue of taking perceptual experience to have imagistic content is that it
avoids certain arguments against the strong content view. According to Matthen,
linguistic reports of what was perceived may be theory-laden in mentioning only
how things perceptually look to the perceiver. For this reason, linguistic reports
need not reflect what was really perceived. Unlike the content of linguistic reports,
imagistic content is not theory-laden, as it differs from the content of the sentences
used to describe what we perceive. For example, if an art critic is looking at an
artistic garden, she may describe it as ‘beautiful’. The image itself, however, does
not represent the garden as beautiful. Whether it is beautiful or not requires an
application of a theory about artistic gardens.
According to Matthen, the view that the content of perception is imagistic
may appear to raise independent problems. In imagistic memory objects and fea-
tures are presented as located in the past relative to the perceiver, and in visual
imagery objects and features may not be temporally or spatially located relative to
the perceiver at all. Perceptual experience, on the other hand, represents objects
and features as present here and now. But if the contents of both mental imagery
and perception are imagistic, how do we explain the fact that only perceptual expe-
rience essentially represents objects and features as located relative to the here and
now of the perceiver?
It may be thought that perceptual content contains temporal and spatial re-
lations to the here and now of the perceiver. However, Matthen thinks there are
reasons to deny this. His argument runs as follows:
Shared Content Argument
1. The content of an imagination does not contain relations to the here and
now of the perceiver.
2. But a perceptual experience and an imagination can have a shared image.
Conclusion: The content of perception does not contain relations to the
here and now of the perceiver.
But if the content of perception does not contain relations to the here and now of
the perceiver, then how can perceptual experience represent objects and features
as intimately tied to the here and now of the perceiver? Matthen suggests that
there is a cognitive feeling that marks the image as presented as an image of the
here and now. Similarly, there is a feeling of pastness that accompanies the images
of episodic memory.
Both Tye and Matthen assume that perceptual experience has content and
focus on making a positive proposal about the nature of this content. As we have
seen, it’s not possible to draw any definitive conclusions about whether perceptual
experience has content without saying something about what the content is like, if
indeed perception has content.
Introduction
28
William Lycan: As we have seen, several of the arguments for and against the view
that perception has content make particular assumptions about the constituents of
perceptual content and the role of perception. As such, it doesn’t seem that debates
about whether perception has content can be settled in isolation from debates
about the role of perception and the constituents of perceptual content, if indeed
perceptual experience has content.
The question of the constituents of perceptual content, in turn, depends on
other debates in philosophy of mind. One important question is whether percep-
tion can be meaningfully separated from cognition.
According to the modularity of mind hypothesis, systems involved in pro-
ducing particular mental states or mental abilities are modular (Fodor 1983;
Sperber 1994, 2002; Pinker 1997; Carruthers 2006). Beyond its core premise
of modularly describing what different regions of the brain do, the modularity
hypothesis also states that certain kinds of information are encapsulated from
influences from other regions (Fodor 1983). For example, the Müller-Lyer illu-
sion illustrates that perceptual information is encapsulated from belief influence.
It may be understood and believed that the lines are the same length between
the two images, but nevertheless they continue to perceptually appear to be of
different lengths.
For information to be encapsulated it does not suffice that there are no
top-down influences on producing it; it must also be free of influences from other
modules. So the Müller-Lyer illusion does not demonstrate that perception is
modular under a strict interpretation of ‘modularity’. However, it does suggest that
there is some meaningful distinction to be drawn between perceptual states and
cognitive states.
William Lycan grants a weak version of the modularity hypothesis: Even if
there is a gradual transition from truly perceptual states to truly cognitive states,
he thinks it makes sense to talk about perceptual states as distinct from cognitive
states. But he doesn’t believe there is any way to settle the question of which fea-
tures perceptual states represent. If he is right about this, this might have conse-
quences for the debate about whether perception has content.
After considering and rejecting various ways that one might settle what
perceptual states represent, Lycan turns to Siegel’s argument from phenomenal
contrast (2005). Let E1 be a visual experience of someone who has the ability to
recognize elm trees (expert) and who is looking at an elm tree, and let E2 be the
visual experience of someone who does not have the ability to recognize elm trees
(novice) and who is looking at the same tree in the same viewing conditions. The
expert finds the tree familiar, the novice does not. So there is a difference in the
overall phenomenal character of their (perceptual or non-perceptual) experiences.
Where ‘K-properties’ is short for ‘high-level kind properties’, Siegel’s argument can
be articulated as follows:
Introduction 29
Unlike other strategies, Lycan thinks Siegel’s argument holds some initial
promise. However, he thinks that it ultimately fails to settle the debate. One
problem with Siegel’s argumentative strategy, according to Lycan, is that it seems
to over-generate. As formulated, it is restricted to natural kind properties, but the
very same argument can be used to argue that the constituents of perception in-
clude high-level natural kind properties can also be used to argue that the constit-
uents of perception include artificial kind properties (e.g., being a clock radio),
mental state properties (e.g., being depressed), aesthetic properties (e.g., being
gloomy), moral properties (e.g., being a virtuous agent), personal taste properties
(e.g., being attractive), mathematical properties (e.g., being 64 alternating black
and white squares), and event properties (e.g., being a car crash).
To illustrate consider a six-year-old who has not had any art classes and a
skilled art critic. Let E1 be a visual experience of the skilled art critic who is looking
at Edvard Munch’s painting The Scream, and let E2 be the visual experience of the
six-year-old who is looking at the same painting. The art critic has the recogni-
tional abilities to pick out the painting as being Edvard Munch’s The Scream, being
an oil on cardboard painting and being completed in Oslo in 1893. The child does
not. So there is a difference in the overall phenomenal character of their (percep-
tual or non-perceptual) experiences. By running through the argument, we can
presumably get to the conclusion that the overall difference in phenomenology
between E1 and E2 is a difference with respect to their kind-properties, in this
case being Edvard Munch’s The Scream, being an oil on cardboard painting and
being completed in Oslo in 1893. Siegel’s argument for a moderately liberal view
thus seems equally supportive of an extremely liberal view that grants that we per-
ceive extremely high-level properties, such as being Edvard Munch’s The Scream or
being created in 1893. The main problem with the extremely liberal view is that it
doesn’t seem to allow for a non-trivial distinction between perceptual and cogni-
tive states. But if aesthetic properties and other extremely high-level properties are
Introduction
30
6
Some use ‘constitutive dependence’ and ‘metaphysical supervenience’ synonymously. Here
I take it that Horgan distinguishes between the essence of phenomenal consciousness and what it
depends on metaphysically.
Introduction 31
the object is a picture, that there is no other picture at that general location and
that the object is the cause of the crooked-picture experience.
Horgan extends this framework to color. His phenomenal intentionality proj-
ect requires that there are color properties that do not constitutively depend on
the external environment. This is so, because we are experientially acquainted
with color properties. Even your envatted brain has perceptual experiences of
red tomatoes. However, it is a further question whether these color properties are
also located where they perceptually appear to be located, viz., on the surfaces of
things, and whether these color properties are the constituents of the contents of
our beliefs and judgments.
Horgan follows David Chalmers (2006) in thinking that while the color prop-
erties we are directly acquainted with in experience could have been located on the
surface of things, objects in the actual world do not actually possess these kinds of
primitive color properties. So, our perceptual experiences are non-veridical.
However, Horgan doesn’t think that our beliefs and judgments about colors
and color discourse are false. Beliefs about concrete particulars can come to have
object-dependent contents. Likewise, our beliefs can come to have contents that
involve the grounds of our color experiences. These grounds, Horgan argues, are
dispositional color properties in Locke’s sense.
Though Horgan’s proposal at first glance seems rather radical in the content
direction, it appears to meet some of the concerns voiced by thinkers leaning in
the other direction. One of Charles Travis’ concerns about representational views
is that they leave the agent out of the equation. Travis should be less concerned
about an approach along these lines. According to Horgan, the most fundamental
kind of phenomenology is agent-centered. All representation derives from this
kind of intentionality, as far as Horgan is concerned.
Horgan’s view offers a potential reconciliation with some of the opponents of
the content views. The phenomenal intentionality thesis takes the agent to be a cen-
tral source of intentionality and offers an explanation of how this agent-centered
intentionality may be a source of both perceptual content and thought content. The
main objection from the Travis camp to this sort of approach probably wouldn’t be
that it construes representation independently of the agent who is doing the repre-
senting but rather that it treats phenomenology as representational independently
of any higher-order epistemic states.
Tomasz Budek and Katalin Farkas: Like most of the literature on perception,
the majority of chapters in this volume have focused on visual perception. In their
contribution to this volume, however, Tomasz Budek and Katalin Farkas con-
sider the more general question of the nature of the constituents of perceptual
experience in different sense modalities. Their main conclusion is that there is no
uniform answer to the question of what the objects of perception are, not even
within a given sense modality. Among the sense modalities discussed by Budek
and Farkas is the modality of olfaction. Budek and Farkas argue that the object
of olfaction sometimes is a chemical and sometimes is a substance or a thing.
Introduction
32
For example, they think that when we walk into a room in which coffee is being
brewed and take a big sniff, we can correctly be said to have smelled coffee odor
in some circumstances and coffee in others. In the envisaged case coffee odor and
coffee are different elements in the causal chain leading to the experience.
This observation is particularly interesting with respect to the debate about
perceptual content because the identity of the object of perception will determine
the accuracy conditions for the experience. If I have a perceptual experience of
coffee odor, then my experience can be accurate even if there is no coffee in the
room but only a chemist experimenting with odor chemicals. If, on the other
hand, I have a perceptual experience of coffee, then my experience can be accurate
only if there is coffee in the room. If, indeed, the accuracy conditions of perceptual
experience reflect the contents of perception, then my sniffing the very same odor
molecules in the same narrow (or internal) circumstances can amount to percep-
tual experiences with different contents.
Olfaction (and other sense modalities such as touch), Budek and Farkas
argue, differ in this respect from visual perception. While it’s plausible that the
objects of olfaction sometimes are odor chemicals, it is plainly implausible that
the object of visual perception sometimes is light absorbed by the cones of the eye.
That is simply not what we mean by the term of art ‘object of perception’. As the
expression is ordinarily used in the perception literature, an object of perception
has to be phenomenally present to the subject, which the light absorbed by the
cones of the eye clearly is not.
One could object to Budek and Farkas’ proposal that an element in the causal
chain leading to smells sometimes serves as an object of olfaction that it focuses too
much on causation. Just as light is not an object of visual perception, so odor chemi-
cals may not be objects of olfaction. Perceived odors, like perceived color properties,
may be largely brain constructs. Further phenomenological and empirical studies
would be needed to settle this question.
Budek and Farkas do not take a direct stance on the content debate. However,
their contribution raises the question of whether the heavy focus on vision in the
literature on perception is partially responsible for the resistance to the idea that
perception has content.
Consider, again, the case of blind individuals who have been taught to navigate
and detect objects via echolocation. A good number of these individuals develop
visual imagery that veridically represents objects in the individual’s environment. In
this case, the proximal causes outside the head are sound-related. But it is far-fetched
to think that these sound-related proximal causes are the objects of the visual experi-
ence. A more plausible assumption is that the visual experiences of these blind indi-
viduals represent visible features, making the weak content view seem appealing.7
7
I am grateful to Susanna Siegel and an anonymous reader from Oxford University Press for
detailed comments on an earlier draft of this paper.
Introduction 33
References
Content Views
2
2.1. Anti-Representationalism
2.1.1. ENACTIVISM
The main enactivist claim is that we have all the information we need in order
to cope with our environment in the world out there. So we do not need to con-
struct representations at all and, more specifically, we do not need perceptual
1
Enactivism and relationalism often combine, see Noë (2004) and Hellie (forthcoming). 39
Content Views
40
representations either. As Dana Ballard put it: “The world is the repository of the
information needed to act. With respect to the observer, it is stored ‘out there’,
and by implication not represented internally in some mental state that exists
separately from the stimulus” (Ballard 1996, p. 111; see also Brooks 1991; Ramsey
2007).
In short, perception is an active and dynamic process between the agent
and the environment and this dynamic interaction doesn’t have to be (or maybe
couldn’t even be) mediated by static entities like representations (Chemero 2009;
Port & Van Gelden 1995). Another version of the positive claims that enactivism
makes is the following: when we see a scene, it is not the case that the whole scene
in all its details is coded in our perceptual system. Only small portions of it are: the
ones we are attending to. The details of the rest of the scene are not coded at all, but
they are available to us all along—we just have to look (O’Regan 1992; Noë 2004,
esp. pp. 22–24).2
The enactivist version of anti-representationalism covers a wide range of views
that differ from one another in important ways: behavior-based AI, Gibsonian
ecological psychology (Gibson 1966, 1979; Chemero 2009), embodied and dis-
tributed cognition (Hutto & Myin 2013), dynamical systems theory (Port & Van
Gelden 1995), and non-classical connectionism (Ramsey 2007), just to name a few.
I will lump them together nonetheless under the label of enactivism as the objec-
tion I will raise applies to all of them as they all share the premise that there are no
perceptual representations.
2.1.2. RELATIONALISM
2
One may be tempted to point out that these two claims, even if they are true, may only give
us reason to conclude that perceptual representations are not static or not detailed, but they give us
no reason to give up perceptual representations per se. My aim here is not to criticize the arguments
in favor of various versions of anti-representationalism, but to raise a general problem for all
anti-representationalist accounts.
Empirical Problems with Anti-Representationalism 41
The relationalists say that there is a relation between the perceiver and the token
perceived object (as well as its properties: a and F) (Campbell 2002; Martin 2006;
Brewer 2006). The enactivists use a variety of metaphors: what happens when I see
a as F is that I fixate on a’s property F (Ballard 1996). Yet another alternative would
be to say that I pick up a’s F-ness in my ambient optic array (Gibson 1966, 1979;
Chemero 2009).
Some of these positive suggestions of anti-representationalism may be more
promising than others, but I will argue that there are empirical problems with the
very idea of disposing of perceptual representations: it is inconsistent with empir-
ical findings about dorsal perception and about the multimodality of perception.
I will analyze these two problems in the next two sections.
gives rise to conceptual content (see Clark 2001 for a summary on the literature on
this). I do not need to take sides in either of these questions.
In healthy humans, the way the dorsal and the ventral stream works can come
apart in some circumstances, as in the case of the three-dimensional Ebbinghaus
illusion. The two-dimensional Ebbinghaus illusion is a simple optical illusion.
A circle that is surrounded by smaller circles looks larger than a circle of the same
size that is surrounded by larger circles. The three-dimensional Ebbinghaus illu-
sion reproduces this illusion in space: a poker chip surrounded by smaller poker
chips appears to be larger than a poker chip of the same diameter surrounded by
larger ones. The surprising finding is that although our judgment and experience
of the comparative size of these two chips is incorrect as we judge the first chip
to be larger than the second one, if we are asked to pick up one of the chips, our
grip-size is influenced by the illusion in a much smaller degree (Aglioti et al. 1995;
cf. Gilliam 1998; Franz et al. 2003; Haffenden & Goodale 1998). The usual way of
explaining this finding is that our dorsal stream is not fooled (or, more precisely,
only mildly fooled) by the illusion but our ventral stream is.
The same results can be reproduced in the case of other optical illusions. In
the Müller-Lyer illusion, while we (mistakenly) see the two lines as having different
length, our eye- and pointing movements represent them (correctly) as being the
same (Goodale & Humphrey 1998; Gentilucci et al. 1996; Daprati & Gentilucci
1997; Bruno 2001). Similarly, in the case of the ‘Kanizsa compression illusion’ and
the ‘hollow face illusion’, our perception is deceived but our action is much less so
(Bruno & Bernardis 2002; and Króliczak et al. 2006, respectively). Thus, some-
times our ventral visual subsystem attributes a different property to an object from
the one the dorsal subsystem does.
This is the representationalist way of describing the three-dimensional
Ebbinghaus case: we have two perceptual representations, a dorsal and a ventral
one, and they represent the chip as having different size properties. But what can
the anti-representationalist say? If perception is a relation between the perceiver
and the perceived token object’s properties, then we have one perceptual relation
here: the one between the perceiver and the perceived token poker chip. But then
which property of the perceived object constitutes the other one of the two relata
of this relation? The property we experience the chip as having or the one that our
grip-size seems to be tracking? These two perceptual episodes are both relations
to the very same token object: the same poker chip and the properties of this same
poker chip. And two different perceptual episodes cannot be constituted by the
very same perceptual relation.
If, on the other hand, as the enactivist says, “the world is our external mem-
ory,” then what serves as our external memory here: the property we experience
the chip as having or the one that our grip-size seems to be tracking? It is difficult
to see what would even be meant by having two different ‘worlds as our external
memory.’
Content Views
44
coded in x in such a way as to make y available to other systems. The more recent
concept of information-carrying (à la Dretske 1995) will not do either as the differ-
ence between information-carrying and representation according to Dretske 1995
is supposed to be that representations have the function to carry information. But
regardless of how we interpret the concept of function in this definition (Millikan
1984; Neander 1991; Bigelow and Pargetter 1987; Walsh 1996; Nanay 2010b, 2012a),
our perceptual system does seem to have the function to carry information about
the size of the chip that would then help us to approach it with the right grip size.
But then it would follow that the perceptual system does represent this property.
In short, appeal to the distinction between information-carrying and representing
does not seem to help the anti-representationalist.
The second, even more serious problem with this anti-representationalist re-
sponse is that if one takes this route, it seems inevitable that she needs to deny that
perception can be unconscious. If what is constitutive of perception is, rather than
perceptual representations, conscious fixation or a relation between an object and
a conscious experience, then perception must be conscious by definition. But this
is a dangerous conclusion to draw: there seems to be a lot of examples of uncon-
scious perception, from visual agnosia and neglect patients to subliminal priming
and blindsight.
Some anti-representationalists will undoubtedly bite the bullet and embrace
the idea that perception is necessarily conscious, but then they have to give us a
way of analyzing those perceptually guided actions that, like the grasping of the
chip in the Ebbinghaus case or like the goal-directed actions of visual agnosia
patients, are not guided by consciously experienced properties of objects. But this
would go against the consensus in cognitive science, where it is generally assumed
that these episodes are unconscious.
There is a lot of recent evidence that multimodal perception is the norm and not
the exception—our sense modalities interact in a variety of ways (see Spence &
Driver 2004 for a summary and O’Callaghan 2008, forthcoming for philosoph-
ical overviews). Information in one sense modality can influence the information
processing in another sense modality at a very early stage of perceptual processing
(often in the primary visual cortex in the case of vision, e.g., see Watkins et al.
2006). A simple example for this is ventriloquism, where vision influences our
audition: we experience the voices as coming from the dummy and not from the
ventriloquist (see Bertelson 1999). But there are more surprising examples: if there
is a flash in your visual scene and you hear two beeps while the flash lasts, you ex-
perience it as two flashes (Shams et al. 2000).
What is the most important for us from this literature is that the multimodal-
ity of perception presupposes that information from two different sense modalities
Content Views
46
is unified in a shared framework (see, e.g., Vroomen et al. 2001; Bertelson and de
Gelder 2004). Noise coming from above and from the left and visual information
from the upper left corner of my visual field are interpreted by the perceptual sys-
tem as belonging to (or bound to) the same sensory individual (whatever that may
be). This is easy for the representationalist to analyze: vision attributes a property
to a part of the perceived scene and audition attributes a different property to the
same perceived scene. The two different sense modalities represent the same scene
as having different properties.
To put it very simply, multimodal perception seems to require matching
two representations, a visual and the auditory one. If we cannot talk about
perceptual representation, how can we talk about what is being matched? The
auditory sense modality gives us a soundscape and vision gives us a visual
scene and our perceptual system puts the two together. It is difficult to explain
this without any appeal to representations. The enactivist arsenal seems insuf-
ficient: they can appeal to the active exploration of the multimodal environ-
ment, but this is unlikely to help here: we are actively exploring the world that
is given to us in both sense modalities—but this in itself requires multimodal
integration. In short, the active exploration of the environment presupposes
multimodal integration, which, in turn, seems to presuppose representations.
Enactivists could insist that the active exploration of the environment happens
separately in each sense modality—but this is in conflict with the findings
about multimodal integration very early in perceptual processing (as early as
the primary visual cortex, see Watking et al. 2006).
The relationalist version of anti-representationalism also seems powerless as
the relation between the perceiver and the token perceived object that constitutes
perception seems to be the outcome of this process of unifying multimodal in-
formation: our experience of the perceived token object (thus, presumably, the
perceptual relation) is brought about by this unification process. The argument
from multimodality seems to show that the phenomena anti-representationalists
emphasize, be it the active and dynamic exploration of the environment or the
relation to a token object, presuppose the coordination of information in the dif-
ferent sense modalities, but this can only be accounted for in representational
terms.
The anti-representationalist has a further option: they can bite the bullet
and admit that there are sub-personal perceptual representations, but when
it comes to the personal level, there aren’t any. The relationalist version of
anti-representationalism may find this response more palatable than the enactiv-
ists, who are often explicit about not limiting their attention to personal level phe-
nomenon (see esp. Ballard 1996; Noë 2004, pp. 28–32). But John McDowell, for
example, explicitly argued that while a representationalist picture is the correct
one for the sub-personal level, we should accept Gibson’s claims with regards to
the personal level, which would make his view (at least in this respect) a version of
enactivism (McDowell 1994).
Empirical Problems with Anti-Representationalism 47
2.4. Conclusion
I argued that in spite of recent efforts to exile the concept of representation from
the discussion of perception, there are empirical reasons why we should hold onto
this concept: we are unlikely to be able to account for and explain dorsal vision and
the multimodality of perception. If this is true, then we should keep perceptual
representations as one of the most important concepts in philosophy of percep-
tion. The real question then is not whether there are perceptual representations
but what kind of representations they are.
References
Aglioti, S., DeSouza, J. F. X., & Goodale, M. A. (1995). Size-contrast illusions deceive the eye
but not the hand. Current Biology, 5, 679–685.
Ballard, D. H. (1996). On the function of visual representation. In Kathleen Akins (Ed.),
Perception (pp. 111–131). New York: Oxford University Press.
Bermúdez, Jose Luis (2000). Personal and subpersonal: A difference without a distinction.
Philosophical Explorations, 3(1), 63–82.
Content Views
48
Króliczak, Grzegorz, Heard, Priscilla, Goodale, Melvyn A., & Gregory, Richard L. (2006).
Dissociation of perception and action unmasked by the hollow-face illusion. Brain
Research, 1080, 9–16.
Martin, M. G. F. (2004). The limits of self-awareness. Philosophical Studies, 120, 37–89.
Martin, M. G. F. (2006). On being alienated. In T. Gendler & J. Hawthorne (Eds.), Perceptual
Experience (pp. 354–410). Oxford: Oxford University Press.
McDowell, J. (1994). The content of perceptual experience. Philosophical Quarterly, 44,
190–205.
Millikan, Ruth G. (1984). Language, Thought and Other Biological Categories. Cambridge,
MA: MIT Press.
Nanay, Bence (2010a). Attention and perceptual content. Analysis, 70, 263–270.
Nanay, Bence (2010b). A modal theory of function. Journal of Philosophy, 107, 91–109.
Nanay, Bence (2012a). Function attribution depends on the explanatory context. Journal of
Philosophy, 109, 623–627.
Nanay, Bence (2012b). Perceiving tropes. Erkenntnis, 77, 1–14.
Nanay, Bence (2013). Between Perception and Action. Oxford: Oxford University Press.
Nanay, Bence (2014). The representationalism versus relationalism debate: Explanatory
contextualism about perception. European Journal of Philosophy, in press.
Neander, Karen (1991). Functions as selected effects. Philosophy of Science, 58, 168–184.
Noë, Alva (2004). Action in Perception. Cambridge, MA: MIT Press.
O’Callaghan, C. (2008). Seeing what you hear: Crossmodal illusions and perception.
Philosophical Issues, 18, 316–338.
O’Callaghan, C. (forthcoming). Perception and multimodality. In E. Margolis, R. Samuels, &
S. Stich (Eds.), Oxford Handbook to Philosophy and Cognitive Science. Oxford: Oxford
University Press.
O’Regan, K. (1992). Solving the “real” mysteries of visual perception: The world as an out-
side memory. Canadian Journal of Psychology, 46, 461–488.
O’Regan, K. (2011). Why Red Doesn’t Sound Like a Bell. New York: Oxford University Press.
Pautz, Adam (2010). An argument for the intentional view of visual experience. In Bence
Nanay (Ed.), Perceiving the World (pp. 254–309). Oxford: Oxford University Press.
Peacocke, Christopher (1992). A Study of Concepts. Cambridge, MA: MIT Press.
Port, R., & van Gelder, T. J. (Eds.) (1995). Mind as Motion: Explorations in the Dynamics of
Cognition. Cambridge, MA: MIT Press.
Ramsey, W. M. (2007). Representation Reconsidered. Cambridge: Cambridge University
Press.
Shams, L., Kamitani, Y., & Shimojo, S. (2000). What you see is what you hear. Nature, 408,
788.
Siegel, Susanna (2010a). Do visual experiences have contents? In Bence Nanay (Ed.),
Perceiving the World (pp. 333–368). Oxford: Oxford University Press.
Siegel, Susanna (2010b). The Contents of Visual Experience. New York: Oxford University
Press.
Soteriou, M. (2000). The particularity of visual perception. European Journal of Philosophy,
8, 173–189.
Spence, C., & Driver, J. (Eds.) (2004). Crossmodal Space and Crossmodal Attention.
Oxford: Oxford University Press.
Travis, C. (2004). The silence of the senses. Mind, 113, 57–94.
Content Views
50
Vroomen, J., Bertelson, P., & de Gelder, B. (2001). Auditory-visual spatial interac-
tions: Automatic versus intentional components. In B. de Gelder, E. de Haan, & C.
Heywood (Eds.), Out of Mind (pp. 140–150). Oxford: Oxford University Press.
Walsh, D. M. (1996). Fitness and function. British Journal for the Philosophy of Science, 47,
553–574.
Watkins, S., Shams, L., Tanaka, S., Haynes, J. D., & Rees, G. (2006). Sound alters activity in
human V1 in association with illusory visual perception. NeuroImage, 31, 1247–1256.
3
A hole in the ground protects some creatures but endangers others. Dry ground
is passable by creatures who walk, but fatal for a fish. These environments provide
different possibilities for different creatures.
J. J. Gibson invented the word “affordance” to denote possibilities of action
for a creature that are given by the environment.1 He proposed that we perceive
affordances, and that the paradigmatic perceptions are byproducts of action plans.
These proposals inspired an “action-first” approach to visual perception, which
foregrounds the role of the perceiver as an actor.
The action-first approach to visual perception can be contrasted with the
“spectator-first” approach, which foregrounds the role of the perceiver as an
observer. This approach is heir to David Marr’s computational theory of vision,
and like Marr’s theory, it gives a central role in perception to belief-like repre-
sentations. Here the paradigmatic perceptions are observations of scenes with
which one does not necessarily interact, such as watching a sunset.
In recent years, both of these approaches have been used to investigate the
nature of perceptual experience, leading to a divide over the centrality of repre-
sentation in analyzing perception.2 On the surface, the two approaches are easily
reconciled by the hypothesis that affordances are on par with color and shape as
properties represented in experience.3 But even if affordances could in principle
be represented in experience, it is reasonable to ask whether they have to be so
For extensive discussion and criticism, thanks to audiences at Harvard, Rice, Geneva, Charleston,
Madison, Oslo, Glasgow, Copenhagen, Princeton, and Universidad Autónoma de México, as well as to
John Bengson, Ned Block, Berit Brogaard, Alex Byrne, Jeremy Dolan, Anya Farennikova, Grace Helton,
Sheridan Hough, Zoe Jenkin, Sean Kelly, Fiona Macpherson, Eric Mandelbaum, Farid Masrour, Laura
Perez, Álvaro Peláez, Sebastian Rödl, and Charles Siewert, and to Miguel-Angel Sebastian and Anna
Bergqvist for writing comments on several drafts. Thanks most of all to Sebastian Watzl for many
illuminating conversations about every aspect of this material.
1
Gibson 1977.
2
Proponents of the action-first approach include Hurley (1998), Noë (2006), Kelly (2006, 2010),
Orlandi (forthcoming). Proponents of spectator-first approach include Byrne (2009), Chalmers (2005),
Pautz (2010), Peacocke (1995), Siegel (2010).
3
For developments of this idea, see Bengson, n.d. and Nanay (2011). 51
Content Views
52
4
Kelly (2006) suggests that experiences of color and shape are always experienced mandates, in
which colors and shapes solicit us to view them in certain ways, and that none of these experiences
involve representation.
5
Merleau-Ponty (1962); Kelly (2005); Cussins [1990] (2003); Dreyfus (2002), (2005).
Affordances and the Contents of Perception 53
6
We can also be moved to action by unconscious perception of things. In a broader sense than the
one used here, those perceptions would be perceptions of soliciting affordances as well.
7
I’ve called these affordances “soliciting affordances” and “non-soliciting affordances,” and it is
natural to call the corresponding experiences “soliciting experiences” and “non-soliciting experiences.”
But please disregard any suggestion by these locutions that the affordances or the experiences are (or
are felt as) the things that do the soliciting. According to a soliciting experience, the bearer of the
affordance solicits you to phi, and according to a non-soliciting experience, something affords phi-ing.
Affordances and the Contents of Perception 55
soliciting entrance on one occasion but not soliciting entrance on another, even if
it is experienced as proto-affording shelter both times.
Among soliciting experiences, we could distinguish between increments of
felt solicitation. A piece of cake might look perfectly positioned to be eaten, but
not so appealing that you experience it as something that is calling out to be eaten.
The hair over the eye might look easily movable, but you might have reconciled
yourself to the hairstyle, so that you don’t experience the hair as calling loudly to
be moved aside. These are differences in increments of felt solicitation.
Experienced mandates belong to the category of soliciting experiences and
have a high degree of felt solicitation. With the ball coming toward you in a tennis
game, the felt solicitation to swing your racket and hit it might be so strong that no
other option enters your mind.
But what exactly is it to feel solicited to do something (to phi) by a slice of
cake, a forest, or a tennis ball? In particular, how does this feeling relate to a moti-
vation or urge to phi? Conceptually, solicitation in the sense used here and moti-
vation can come apart. Suppose you hear music that (you can tell) is designed to be
danced to, but you feel completely unmoved by it to dance.8 The music, as you hear
it, is telling you to dance, but you don’t feel its pull. It is as if the music is trying to
make you dance, but you are not cooperating and have no inclination to cooperate.
Similarly, the neatly stacked colorful packs of candy thoughtfully placed at eye
level in the grocery store check-out line are supposed to make us disposed to buy
them. The exhortative tone of many advertisements on radio and television has an
analog in visual displays, and we can feel that they are designed to propel us into
purchasing the things advertised, even when we feel not at all moved to buy the
allegedly indispensable item. It might seem natural to call any experience of ex-
hortation an experienced mandate. If seeing a piece of cake moves you to want to
eat it when you weren’t even hungry to begin with, you are mobilized from scratch
in just the way that advertisers dream of. But advertisers’ coercive intentions to
generate desires are often detectable, even when the coercion does not succeed. I
reserve the label “experienced mandate” for soliciting experiences involving some
increment of motivation to do what is solicited.
The phenomenal aspect of experienced mandates whereby they are motivated
constitutes what I’ll call a feeling of answerability. This feeling also normally results
from hearing one’s name called or from meeting another person’s gaze. (Given
the developmental importance of joint attention, we should expect special sensi-
tivity to the direction of other people’s gaze.) We say that a person answers to one
name but not to another (e.g., “Julia” but not “Julie”). To “feel answerable to ‘Julia’ ”
means not that one can answer to “Julia,” but that one does answer to Julia.
Like answerability itself, the feeling of answerability is structurally similar to
responsibility: you can shirk a responsibility, but you won’t thereby cease to be
8
I thank Farid Masrour for this example.
Content Views
56
9
Bortolotti 2013.
Affordances and the Contents of Perception 57
in that situation and does not absorb your perceptual attention.10 Whereas with
the dance music you were solicited to do something but not motivated to do it,
here you are motivated to something, without anything in your immediate envi-
ronment soliciting you to do it. Experienced mandates involve affordances by the
perceived environment that you’re currently in. In experienced mandates, solicita-
tion and motivation go together.
A final observation about experienced mandates is that the thing that one
experiences as soliciting can either be localized in an object (cake, hair), or it can be
an entire situation. Suppose you are alone on a narrow sidewalk, walking. The side-
walk affords following its path. Then someone turns a corner and begins walking
toward you. They are still far off, and no one else is in between. In this completely
ordinary situation, without having to think about it all, you assume that the person
is going to continue walking toward you until you pass. And now the affordance
of travel is more complicated. When your paths cross, the part of the sidewalk tra-
versed by them will not be passable. But you don’t yet know exactly which part
this will be. The space has to be negotiated by adjusting your relative positions.
Traversing the sidewalk’s path is afforded, but it is not afforded simply by the side-
walk. It is afforded by the sidewalk together with the passerby, contingent on their
cooperation. It remains to be seen what form the cooperation will take. The same
possibilities are open to both of you: step aside to let the other pass, or continue on
whatever path they open up for you? Make clear gestures designed to acknowledge
the other person, or play down the fact that there’s any interaction? Acknowledge
any adjustment they make, or just carry on? Besides these possible modes of full
cooperation, there is also the possibility of grudging cooperation, borderline
non-cooperation (barely move out of the way), or at an extreme, collision.11
So experienced mandates are motivated, soliciting experiences of affordances
of things in the immediate environment, including entire situations. These are the
experiences that our questions Q1‒Q3 are about.
We can distinguish between three temporal relationships an experience can
bear to the mandated action that it presents. First, one might experience as man-
datory an action not yet undertaken. Second, one might experience as mandatory
an action now being completed as mandated. Third, one might retrospectively
experience as having been mandated an action just completed.
10
An experience like the one described could unfold in the subject’s immediate environment,
if the subject was in an agitated state in which everything in the plane came to look like an obstacle
to making the crucial phone call. Perhaps such a subject would experience their surroundings in the
plane as anti-affording cancelling their electricity service. (In principle, though probably not in fact,
a subject’s agitation could even make oblong items start to look to them like phones). But to create an
illuminating foil for experienced mandates, in the plane example I’m assuming that the subject is not
even experiencing the environment as either affording or anti-affording the task of cancelling their
electricity service.
11
These possibilities are made vivid in T. G. Seuss’ story The Zax, in which a north-going Zax and
a south-going Zax are stopped in their tracks because neither will move out of the other’s way.
Content Views
58
In experience (i), you experience the mandate without executing the mandated
action right then and there—before executing it, or without ever executing it.12 In
principle, a situation could present the same affordance as mandated, first in way
(i) and then subsequently in ways (ii) and (iii). But we shouldn’t assimilate (i) to
an anticipation of executing the action. Just as you might receive a mandate from
someone else that you go on to ignore, so too you might experience a potential
action (such as moving the hair aside) as mandatory in way (i), without going on
to execute it, though possibly only at the cost of some dissonance.
I’m going to use “experienced mandates” to denote dynamic experiences
spanning types (i)‒(iii). But in discussing how experienced mandates relate to rep-
resentation, I’ll focus mainly on experiences of type (i).
12
Kelly (2006) suggests that experiences characterized in (i) are typical in perceiving shape and
color constancies. In perceiving the color or shape of something, he argues, we are sensitive to an
optimal point from which it could be viewed. He does not claim, however, that every experience of
perceptual constancies is one in which we actually optimize our bodily position vis-à-vis the thing
we’re seeing.
Affordances and the Contents of Perception 59
13
The notion of tension comes from Merleau-Ponty, who uses it in Phenomenology of Perception
to describe the phenomenon of perceptual constancies. He writes “The distance from me to the object
is not a size which increases or decreases, but a tension which fluctuates around a norm.” Quoted in
Kelly 2010.
Content Views
60
game, what one experiences is more like one’s arm going up and its being
drawn to the appropriate position, the racket forming the optimal angle
with the court—an angle one need not even be aware of—all this so as
to complete the gestalt made up of the court, one’s running opponent,
and the oncoming ball. One feels that one’s comportment was caused by
the perceived conditions in such a way as to reduce a sense of deviation
from some satisfactory gestalt. But that final gestalt need not be repre-
sented in one’s mind. Indeed, it is not something one could represent. One
only senses when one is getting closer or further away from the optimum.
(2002, p. 379)
Now, much of what we do is in a broad sense skilled action—a fact easily observed
by watching toddlers in the midst of learning how to open doors or put on their
shoes. In contrast, Dreyfus focuses on specialized skilled action (tennis), rather
than on sensory-motor habits that nearly everyone eventually develops (tying
shoes, opening doors). But experienced mandates can be found in the domain of
sensory-motor habits as well. In the passerby example, one can imagine feeling
that stepping aside to let the other person pass is mandatory, in the sense that
among the possible modes of passing another person on the path, moving aside
to let the other person pass is the only one experienced as possible, and in that
sense, mandatory. The type of mandate can vary with the situation. Perhaps the
passerby is frail, or moves only with difficulty, and the felt mandate stems from
moral sensitivity.14 Or perhaps one has cultivated a habit of always letting the
other person pass, out of politeness, or because one enjoys determining how
such micro-interactions with the public unfold. Here too, the mandatory aspect
stems not from specialized motor skill, but from a broadly social sensitivity. In
yet other cases the felt mandate might be broadly aesthetic, as it is in the example
involving the tuft of your interlocutor’s hair that falls just in front of their left
eye, making it harder to read their expression, and producing in you a strong
impulse to move the hair out of the way. Or in an exhausted state, a fluffy bed in
an empty room might be experienced as inviting you to plop down on it for rest.
The felt mandate does not come from specialized skilled action or its dynamics
of execution in any of these examples.
The challenge posed by experienced mandates to the primacy of represen-
tation is potentially quite powerful. Experienced mandates pervade much of our
conscious lives, arising both in habitual action and in specialized skilled action.
In light of the broad challenge, let us ask: what is the relationship between expe-
rienced mandates and representation?
14
In an excellent discussion of similar phenomena, Bengson n.d., following Mandelbaum (1969)
discusses an example of this sort, where someone gives up their seat on the bus to someone else who is
visibly tired. A similar example occurs in Murdoch (1958).
Affordances and the Contents of Perception 61
We can begin our inquiry into this relationship by asking whether experienced man-
dates are experiences with accuracy conditions, where the accuracy conditions char-
acterize how the environment seems to the subject to be. If an experience has accuracy
conditions, then it is accurate only if those conditions are satisfied. When questions
Q1‒Q3 ask about the role of representational contents in experienced mandates, they
are asking about the kind of accuracy condition that would characterize the experi-
ence from the subject’s point of view. Discerning just which accuracy conditions do
that is no small task.15 But some candidates naturally suggest themselves. For instance,
if an experience of a tennis ball hurtling toward you has an accuracy condition, it
might include the condition that the ball is green and coming toward you. A prelimi-
nary question, however, is whether experiences have any representational contents at
all. And a version of that question concerns experienced mandates:
Q1. Do experienced mandates have any representational contents?
So far, I have glossed over the exact relationship that experienced mandates bear
to perceptual experience, leaving open that they might simply be perceptual expe-
riences, or alternatively they might have perceptual experiences as components. It
seems obvious that perceptual experiences are related to experienced mandates in
one of these two ways. You couldn’t very well play tennis without seeing the ball, or
seek protection by entering a forest without perceiving where the forest is in relation
to you.
The central observation favoring the idea that perceptual experiences in expe-
rienced mandates are contentless is that the subject feels immediately solicited by
their environment in ways that move them to action. There might then seem to be
no explanatory role for contents of experience to play.
In reply, the fact that while acting easily in an environment, one is seemingly
propelled by it does not undermine the general considerations about perception
that suggest that such experiences have contents. Consider the case of visual expe-
riences, since all of our examples involve vision. According to the Content View,
all visual perceptual experiences have contents. The central motivation for the
Content View is phenomenological. When you see things, they look to you to be a
certain way. And when they look to you to be a certain way, they look to have cer-
tain properties. I won’t repeat a full defense (given elsewhere) of the Content View
here, but the key transition in that defense moves from ‘X looks to have property
F’ to ‘The experience of X’s having F is accurate only if X has F’.16
15
We can’t read off directly from introspection which contents experiences have. For discussion,
see Siegel 2010, ch. 4.
16
Siegel 2010, ch. 2.
Content Views
62
On the starting point of the transition, the fact that Dreyfus’ tennis player can
so easily navigate the tennis court does not entail that the things the player sees on
the court—the player, the oncoming ball, the spaces between himself and net—fail
to look any way to him at all. Without seeing the tennis court, it would be hard to
play tennis and hard to perceive the gestalt to be completed. As Dreyfus observes, in
the experience of playing tennis, there are some “perceived conditions” in response
to which one adjusts one’s movement. Since the adjustments are made to “complete
the gestalt” of the tennis game, these perceived conditions presumably include the
components of the gestalt, such as the positions of the opponent and the ball. Which
ways these things look will depend on many factors, including what you’re doing—
such as whether you’re playing the game, watching the players, or studying the court
in order to draw it. But if the court, or the things in it, or the situation on the court
didn’t look any way to you at all, then you would not be seeing them. (Even if you
are hallucinating rather than seeing, the same basic phenomenological point holds.)
In all of these cases, properties characterize the way things look to us, when we
see them. And if things look to have certain properties, then, it seems, the experience
is accurate, only if things have the properties that they look to have. Of course, posi-
tive reason is needed to think that in general, when things you see look to you to have
certain properties, the experience is accurate only if the things are the way they look.
A full defense of this transition is the core of my case for the Content View.
What’s important here is that nothing specific to experienced mandates forces any
departure from the starting point of this general argument.
Although it may seem obvious from our examples so far that experienced man-
dates involve perceptual experiences, different examples might call this assumption
into question. And if there are experienced mandates that extinguish all percep-
tual experiences as they unfold, then trivially, no contentful perceptual experiences
would play any important role analyzing experienced mandates. That result would
open the possibility of denying that experienced mandates have representational
contents, challenging the significance of the Content View, while tolerating its truth.
One might hypothesize that the extinction of perceptual experiences is a spe-
cial case of an ordinary occurrence, in which we exercise skills or habits without
much guidance, if any, from perceptual experience. So it is worth considering
whether some experienced mandates plausibly extinguish perceptual experiences
we may start out having in earlier stages of habit or skill-formation, but which
fade out completely by the time the habit or skill is well-established. Perhaps in
some such cases, we form beliefs about what’s around us, without basing those
beliefs on any experience. One often doesn’t need to look carefully, or at all, to see
where to reach for a familiar doorknob, because one’s body ‘knows’ already, out
of sensory-motor habit, or thanks to unconscious visual processing in the dorsal
stream.17
17
Milner and Goodale 1995.
Affordances and the Contents of Perception 63
18
Dreyfus 2005, p. 8 of web version.
19
On attentional suppression, see van Rullen and Koch 2003.
Content Views
66
mandates entering it. A third option is neither factor is explanatorily prior to the
other (perhaps they are connected by a feedback loop).
In the content-first and feedback loop options, the contents of the percep-
tual experience could clearly help explain why the subject performs the man-
dated action. First, they contribute to making the affordance salient, and this
helps explain why it is experienced as a mandate. Second, since on these options
the component perceptual experience is explanatorily upstream of the experi-
enced mandate, they help explain the role of the experienced mandate in guiding
action.
What if the rationalizing contents are related to the mandate in the action-first
direction (e.g., the bed looks fluffy, because the subject experiences plopping down on
it as mandated)? The perceiver collapses onto the bed because she is exhausted, not
because of the way the bed looks to her. Here the contentful sub-experience contrib-
utes to the perceiver’s intellectual coherence and integrity. Compare a case of psycho-
logical (as opposed to normative) rationalization. When people with excessive fear
of heights stand on high balconies, their acrophobia ends up exaggerating how high
they believe the balcony to be, compared to height estimates by non-acrophobes.20 Let
us suppose for the sake of argument that they don’t fear the height because the bal-
cony seems so high off the ground, but rather that the balcony seems so high off the
ground in part because they are afraid of heights. On the assumption that with all else
equal, it is more reasonable to be nervous about standing on a higher balcony than
a lower one, the acrophobes’ mistaken belief about how high the balcony is brings
their fear into harmony with beliefs—even if the beliefs themselves are unreasonable,
caused as they are by an excessive fear, rather than by an accurate assessment of the
situation. Even the craziest, most irrational subjects sometimes display this type of
internal cognitive harmony, such as the schizophrenic patient who is highly anxious
because he thinks that the world is about to end and finds the arrangement of chess
pieces on the chessboard to be ominous.
In these cases, the beliefs that the chess pieces are ominous and that the bal-
cony is very high rationalize the background anxiety or fear in something like
the way that the fluffy-bed experience and the contentful sub-experiences in our
other examples rationalize experienced mandates. The contentful states in all of
these cases, whether they are beliefs or experiences, give us a way to describe this
phenomenon in which subjects (or their subpersonal processes) bring their psy-
chological states into a type of cognitive harmony.21 To the extent that contentful
states figure in these processes, they are not explanatorily idle.
20
Stefanucci and Proffitt (2009) provide some evidence that something like this phenomenon
actually occurs. Using a variety of measures, both acrophobes and non- acrophobes tend to overestimate
the height of balconies they are standing on, but acrophobes exaggerate the height substantially more
than non-acrophobes.
21
If perceptual experiences stand in rational relations to one another, then they belong to a
domain in which rational assessment apply. For discussion of larger implications for this idea for
epistemology, see Siegel 2013 and Siegel, forthcoming.
Affordances and the Contents of Perception 67
Let us turn to the third question about the role of representation in experi-
enced mandates.
Q3. Is there any proposition that could reflect the mandate that the subject
experiences, if that proposition were the content of the experience?
If all we knew about a subject’s perceptual experience is that it had non-rationalizing
properties, then we could not read off from her perceptual experience the fact that
a subject feels solicited to do something by her environment, let alone that she is
motivated to meet the situation with the solicited action. Neither could we read
this off from the fact that the experience represents properties that would ration-
alize plopping down on a bed or moving a tuft of hair out of the way. Some other
aspect of the experience would have to account for the soliciting and the motivat-
ing aspects of experienced mandates.
Can we identify a proposition that could reflect both the soliciting and moti-
vating aspects of experienced mandates? Let us start with the soliciting aspect of
experienced mandates. How could contents locate the solicitation in something
perceptually experienced? Contents of the form “X is to-be-phi’d,” such as ‘the
hair is to-be-moved’, ‘the passerby is to-be-made-way-for’ are a straightforward
way to characterize the mandated action, while preserving the idea that the man-
date is experienced as issued by something external that we perceive. Consider
the difference between the predicates “to-be-phi’d” and “to-be-done.” Whereas
“to-be-moved-out-of-the-way” is applied to the interlocutor’s tuft of hair or to
some other obstacle, “to-be-done” is applied to actions, such as moving hair out
of the way, or making room on the sidewalk. Given the assumption that solicita-
tions are experienced as being issued by things like cake and forests, a natural
hypothesis is that experienced mandates represent specific to-be-phi’d properties,
attributing them to things perceptually experienced.
What about the motivating aspect of experienced mandates? To bring this
aspect of experienced mandates back into focus, recall the two music cases. The
two experiences of music both solicit dancing, but they differ in whether they
motivate the perceiver to dance. What, if anything, can differentiate between
pairs of cases that differ only in this way, consistently with experiencing the
mandate as issuing from the thing perceived? Could the motivational part of the
mandate be reflected in contents that characterize how the music sounds, or how
the forest, hair, sidewalk, tennis ball, etc. look?
One might think this aspect of experienced mandates is not representable in
experience, on the grounds that it is a conative state, and experiences only take a
stand on how the world actually is, not on how one wants it to be or is motivated
to shape it. If Hume, Searle, and others who endorse this sharp division are right,
then at most, the soliciting aspect of experienced mandates could be represented
in experience.
A first attempt to find a place for the motivating aspect of experienced man-
dates in accuracy conditions adds the issuing of a mandate by X (cake, forest, hair,
Content Views
68
etc.) to the ‘to-be-phi’d property. The result would be that experience has conjunc-
tion of contents. Here are some candidates for the conjunction:
Exp: X is to-be-phi’d and X . . .
. . . wants me to phi
. . . is telling me to phi
. . . commands me to phi
. . . intends for me to phi.
But this option does not identify any difference between the two music cases,
for the same reason that ‘X is to-be-phi’d’ does not identify any such difference.
One need not feel moved by what X wants, tells you, or intends. As evidenced by
long-standing, unfulfilled to-do lists, representing that something is to be done,
for instance by writing it down on a list of things to do, does not suffice to moti-
vate you to do it.
A different strategy is to complicate the second component of content further,
by adding a causal relation that links the soliciting aspects to a desire or an action:
• Exp: X is to-be-phi’d, and the fact that X is to-be-phi’d
. . . makes me want to phi.
. . . is making me phi.
• Exp: X is to-be-phi’d, and because X is to-be-phi’d, I am going to phi.
These proposals posit ascriptions of one’s own desires or intentions as part of the
content of experience. One might worry that this fails to respect the way in which
perceptual experience is directed outward, characterizing things external to the
subject’s mind. The phenomenal integration of solicitation and motivation is re-
flected in the unreflective nature of habitual and specialized skilled action. Sartre
describes the integration when he uses a locution of the form “to-be-phi’d” to
describe being “plunged into the world of objects”:
When I run after a streetcar, when I look at the time, when I am absorbed in
contemplating a portrait, there is no I. There is consciousness of the streetcar-
having-to-be-overtaken . . . I am then plunged into the world of objects; it is
they which constitute the unity of my consciousness; it is they which present
themselves with values, with attractive and repellant qualities—but I have dis-
appeared; I have annihilated myself. There is no place for me on this level.
And this is not a matter of chance, due to a momentary lapse of attention, but
happens because of the very structure of consciousness.22
Sartre may overstate the ‘disappearance of the subject’. Representations of direc-
tionality and distance have an implicit first-person component, so contents involv-
ing the subject seem indispensable. But he seems right that the subject disappears
as a subject of desire. We need not be aware of our motivation to hit the tennis ball,
22
Sartre 1957, pp. 48‒49. I put in bold the locution of interest.
Affordances and the Contents of Perception 69
or flick the light switch, or move aside to let the oncoming pedestrian pass. We’re
just aware of the ball, the switch, the sidewalk, etc.
A more promising proposal is that the motivating aspect of experienced man-
date casts its shadow on the contents of experience, in the form of contents that
include a property related to answerability. What’s needed is a hypothesis that
identifies contents that are correlated with the feeling of answerability to a solicit-
ing affordance.23 The effort to find such contents may seem futile, for any of three
reasons:
Regarding the first challenge, assuming the notion of a felt valence has psycholog-
ical validity, it seems reasonable to suppose that it figures in feelings of answer-
ability in experienced mandates.24 But it is doubtful that the felt valences exhaust
those feelings of answerability. If a pack of chewing gum looks appealing, it need
not cost you any dissonance to ignore it. Here the perception of the gum does not
involve any persisting inner mobilization of the sort that characterizes the feeling
of answerability.
The other two challenges are not easily rejected. On the face of it, feeling
answerable to a solicitation seems to be a mode of experiencing the solicitation.
One experiences it in a mobilizing way, as opposed to experiencing it indiffer-
ently. This observation suggests that it must belong on the side of the force/
content distinction belonging to force. And while a natural candidate for an ac-
curacy condition associated with the feeling of answerability would seem to be
“A is to-be-answered,” this suggestion invites the charge that like other ‘to-be-
phi’d’ contents, it too can be entertained indifferently.
The strongest answer to these substantial challenges combines structural con-
siderations with closer attention to what is experienced in the feelings of answer-
ability that are at issue. The structural consideration is that contents could have a
nested structure that reflects the main idea driving Challenge 2. The nested struc-
ture seems to respect the fact that there are both motivated and indifferent ways to
23
I leave open whether the contents are constitutively linked to the phenomenal character, rather
than being merely correlated with them.
24
For discussion of microvalences, see Lebrecht et al. 2012.
Content Views
70
experience soliciting affordances. Leaving a crucial element partly blank for now,
the structure could be this:
Experience: [It is < . . . answerability . . . > that: X is to-be-phi’d].
What fills in the ellipsis to create an accuracy condition? “. . . to-be-answered . . . ”
sounds prospective, and suggests that answering to the soliciting affordance is
something the subject may or may not go on to do, whereas what we’re looking
for is a way to reflect the fact that the subject is already answering to the soliciting
affordance. A better proposal is thus:
Experience: [It is answered that: X is to-be-phi’d.]
I’ll call the contents on the right side of the word “Experience” answerability contents.
Of course these contents are not anything one would find natural to say in
describing the experience. But the same is true of many other accuracy conditions.
The proposal here respects the integration of soliciting and motivating aspects
of experienced mandates, and the fact that the soliciting affordance generates a
feeling of answerability.
In cases where an experience also represents rationalizing properties, the
contents embedded in “it is answered that” may be more complex, integrating the
rationalizing properties with the to-be-phi’d property. For instance, “It is answered
that: the bed is to-be-plopped down upon and is fluffy.” I leave it open whether the
rationalizing relation itself might be represented in experience, as would be reflected
in contents such as “ . . . to bed is to-be-plopped down upon because it is fluffy.”25
What are some cases in which these accuracy conditions are met, and what
are some cases in which they are not met? It seems plausible that the feeling of
answerability suffices for the subject of the feeling to be answering to something
in a minimal way—a way that does not consist in taking the thing that they are
answering to (such as a piece of cake) to be a source of normative constraint. If
so, then answerability contents are always correct. Compatibly with this result, a
subject could in principle make an introspective error about whether she is or isn’t
feeling answerable to something.
In addition to probing the conditions under which answerability contents are
true, we can also ask about the conditions under which the ‘X is to-be-phi’d’ contents
they embed are true. What would it take for it to be the case that the hair really is
to-be-moved, or that the forest really is to-be-entered, the oncoming pedestrian is
to-be-made-way-for, or that the cake really is to-be-eaten? These questions have no
general answer, because the contents do not specify what kind of ‘ought’ underlies
the mandate. For instance, if the morally correct thing to do on the path is move
aside by giving the passerby lots of room to pass, then relative to moral ‘ought’, the
pedestrian is to-be-made-way-for. In the hair case, relative to the ‘ought’ of social
mores, it is not the case that the hair is to-be-moved, but perhaps relative to the
25
For discussion of seeing reasons, see Church 2010.
Affordances and the Contents of Perception 71
‘ought’ of communicative efficacy or aesthetic rightness, it is the case that the hair
is to-be-moved (assuming that the interlocutor really does have an eye underneath
the hair).
Even though answerability contents do not specify a norm relative to which
they are accurate, which mandates a person experiences can indicate which norms
she is sensitive to. In talking to a person you might feel a mandate to stop lis-
tening, or deeply discount what they say, or (on the other side) to put a lot of stock
in what they say. Such experiences might manifest and perpetuate background
attitudes of deference or disrespect. They illustrate the potential use of the frame-
work of answerability contents in analyzing the interpersonal interactions. Such
modes of aversion and approach can be the social instrument by which social
patterns are maintained, such as patterns of exclusion and inclusion, or trust and
dismissiveness.
These considerations can illuminate the conditions under which to-be-phi’d
contents would be accurate. Relative to the social norms that such experienced
mandates manifest, the answerability contents would be accurate, where relative
to epistemic or moral norms, they might be inaccurate. For instance, if one experi-
ences one’s interlocutors’ comments as to-be-discounted when such discounting
would be epistemically inappropriate, then the ‘to-be-discounted’ contents could
be accurate relative to a social norm that mandates discounting, but inaccurate
relative to epistemic norms.
The ‘to-be-phi’d’ contents that answerability contents embed are thus not
complete accuracy conditions, because they leave unspecified a parameter that
needs to be fixed in order to generate an accuracy condition. The reason to think
this parameter is left unspecified is that there don’t seem to be phenomenal dif-
ferences that track different norms relative to which to-be-phi’d contents could be
assessed for accuracy.
I’ve replied positively to question Q3 by arguing that answerability contents
could identify a proposition that reflects the mandate that the subject experiences.
Answerability contents are always true, and their embedded contents can be accurate
or inaccurate only once an additional parameter is fixed. In light of these facts, does
this response to Q3 concede anything to the overall challenge to the primacy of rep-
resentation from experienced mandates? To the extent that the embedded contents
are accurate only relative to a type of norm that the experiences do not themselves
specify, experienced mandates fail to provide full accuracy conditions by themselves.26
But rather than being irrelevant to experienced mandates, answerability contents
let us express important features of this phenomenon. They let us identify rational
relationships between subexperiences, they provide a framework for understanding
26
Klein (2007) observes that experiences with imperatival contents (such as “step gingerly on
your left foot”) would not threaten representationalism about experience. This observation is true but
irrelevant to the discussion of experienced mandates, since replacing the declarative content “the hair
is to-be-moved” with the imperatival content “move the hair” would not suffice to reflect the motivated
aspect of experienced mandates.
Content Views
72
With answerability contents on the table, we can return to the stronger thesis that
experienced mandates have contents like these. This thesis goes beyond answering
our initial three questions Q1‒Q3. Can we discover whether experienced man-
dates have such contents?
A possible strategy starts from the substantial assumption that affordances
can be represented in experience (and that experiences therefore have accuracy
conditions), rather than trying to defend the stronger thesis from the ground up.
Given this assumption, one might try to argue for two conditionals:
Affordances → Solicitations: If affordances are represented in experience,
then solicitations are too.
Solicitations → Mandates: If solicitations are represented in experience, then
mandates are too.
Favoring the Affordance → Solicitation conditional, one might reason roughly as
follows. From the cases of advertising and dance music, the feeling of being solic-
ited by the environment to do something is familiar. If we are entitled (by our
starting assumption) to use the idea that affordances are represented in experience
to analyze perceptual experience, it is natural to use it for the special class of salient
affordances that we experience as soliciting. There seems to be no principled bar to
extending the analysis of perceptual experience from representing of affordances
to representing the special case of soliciting affordances.
Favoring the Solicitation → Mandate conditional, one could try to use a
method of phenomenal contrast (Siegel 2010) to evaluate hypotheses about what
best explains of the phenomenal contrast between the two music cases, or another
pair of cases in which two subjects seem to have experiences with the same so-
licitation content, but differ in whether they feel motivated to fulfill the solicita-
tion they experience. The method could either be used to evaluate the hypothesis
that the motivational aspect is best analyzed in terms of accuracy conditions of
the form:
It is answered that: X is to-be-phi’d.
(The same cases and method could also be used to evaluate other hypoth-
eses that posit different accuracy conditions, or something other than accu-
racy conditions.) Since the burden of the strategy is to show that this is the
best explanation of the phenomenal contrast, alternative hypotheses need to
Affordances and the Contents of Perception 73
3.3. Conclusion
Dreyfus and other writers who have described experienced mandates call atten-
tion to an important fact about perception: sometimes our perceptual experiences
are pervasively structured by our role as agents responding to social situations.
Content Views
74
In other situations, our dominant mode is not that of an agent, but a spectator—
for instance when we are freed from immediate pressures of spatial negotiation,
simply taking in our surroundings. These writers are right to emphasize that phe-
nomenologically, perception feels quite different depending on whether it is dom-
inantly structured by our roles as agents or not. And that raises a question: to what
extent are our experiences structured by experienced mandates, to what extent
aren’t they? An upshot of the discussion here is that even if the extent is great, the
role of the spectator never disappears completely, even when we’re in the throes
of action.
References
4.1. Introduction
Whether perceptual experience has content has been a question of lively debate
for the last decade or so. One way of construing the question is whether percep-
tual experiences are mental states much like beliefs and desires—whether they
are mental states, that is, that can be understood on the model of the proposi-
tional attitudes. That is the way in which I shall construe the question in this
chapter. And I shall offer an argument for answering it in the affirmative, an ar-
gument that to my mind at least has not received quite the attention it deserves
in the recent debates. The argument starts from the observation that percep-
tual experience figures in folk-psychological reasons-explanations in a certain
way. And playing certain sorts of folk-psychological roles is arguably one of
the strongest motivations for construing mental states as having content in the
first place—modeling these roles is precisely what contents are for. Whether
experience has content thus clearly is a question of significance not only for the
metaphysics of mind but also for epistemology, the theory of rationality, and the
theory of action.
Asking whether experience has content immediately raises to two further ques-
tions: What are contents? And: What is it for a psychological state to have a con-
tent? These questions are closely related: Asking what it is for a psychological state
to have content amounts to asking what contents are there for, what their theoret-
ical role or function is. And if contents are whatever best fulfills that role, we can
find out what they are by means of investigating what that role is and what best
fulfills it.
In this section, I shall provide a rough-and-ready answer to these questions
and apply it to the question whether experience has content. The account of per-
76 ceptual experience and its content I shall suggest is highly controversial, but it
Looks, Reasons, and Experiences 77
1
Cf. Davidson 1973, 1975, 1974, 2005; Lewis 1972, 1979.
2
There is, I think, a discussion to be had about the theoretical role of the notion of content, a
discussion of some urgency even, especially regarding its relation to the notion of semantic value. But
this is not the right place for that discussion.
3
Davidson, of course, thought that we could do without any objects—be they propositions or
whatever—explicitly assigned as meanings to utterances or contents to attitudes. He not only held
such objects to be ultimately redundant in the theory of meaning and content, but also worried that
proposition-talk might (falsely, according to him) suggest that there are unique correct assignments
of meaning and content (cf. Davidson 1974, p. 147). Nevertheless, Davidson himself talks quite freely
of propositional contents, and the minimal notion of propositional content adopted below should be
entirely compatible with his strictures.
4
I shall remain neutral here on the question of whether this content is structured.
5
Tradition thus has it that all psychological states with contents are propositional attitudes, where
‘propositional attitude’ is taken to have an inclusive and hopefully uncontroversial sense, too. In this
sense, entertainings are propositional attitudes, for instance. Tradition also has it that the attitudes have
uniform contents in the sense that all attitudes have whole propositions as contents. This, I take it, rules
out propositional functions as attitude contents. It does not rule out that the contents of propositional
attitudes are best construed as centered worlds propositions, however. Whether or not they are is
another question I remain neutral on here.
Content Views
78
Contents thus play the following role: They are abstract objects we assign to psy-
chological states to model a specific kind of structure among them, to locate in-
dividual states in such structures, and to predict and explain their formation by
means of other such states present in such structures. The states are those our
folk-psychology recognizes, the explanations are reasons-explanations, and the
individuation of the relevant states is by means of propositional contents and folk-
psychological role. Here is a recent voice, that of Richard Heck, summing up this
way of thinking about content:
Why should we attribute content to mental states at all? A common answer
might be that mental states are representational: Talk of a state’s content is
short for talk of its representational properties. That is certainly true. But why
trouble ourselves with the representational properties of mental states? What
would we lose if we just ignored them? I take it that we would lose the very
idea of psychological explanation. We are in the habit of explaining our own
behavior, and that of other creatures, in terms of what we all believe: We ex-
plain why Joe ran across the room in terms of his believing that his stuffed di-
nosaur was on the other side. . . . The explanations themselves are formulated
not in terms of the neurological features of mental states but in terms of their
contents . . . . And so we might say: The reason we should attribute content to
mental states is because there are things we wish to explain in terms of mental
states, as individuated by their contents. (Heck 2007, pp. 120f.)
If sufficiently fine-grained individuation was our only concern, propositions might
not be the objects of choice for the role of contents; other kinds of abstract objects
might do as well. But psychological explanation requires more than sufficiently
fine-grained individuation; it requires relations beyond sameness and difference.
Propositions by their very nature stand in certain relations to one another that
uniquely qualify them for their job: the logical and inferential or evidential rela-
tions. Heck again:
Why not just take the contents of beliefs to be (possibly transfinite) ordinal
numbers? . . . [T]he best answer, it seems to me, is that mental states are not
Looks, Reasons, and Experiences 79
just distinguished from one another by their contents: They are also related
to one another by their contents. For example, given any two beliefs, there are
several other beliefs that are related to them in familiar ways: Their negations,
their conjunction and disjunction, and so forth. These relations are not just
logical but also psychological: Someone who believes two propositions will,
ceteris paribus, also tend to believe their conjunction, at least when the ques-
tion arises. (Heck 2007, p. 121)
For beliefs, that is, reasons explanations are possible because the beliefs of actual
believers at least to some (minimal) degree tend to instantiate the objective log-
ical and inferential or evidential relations between the propositions that are their
objects. Together with beliefs, the ascription of contents to desires (or pro-attitudes)
then allows for reasons explanation of intentional actions.
From this perspective, content first and foremost is what allows a psycho-
logical state to play a role in reasons-explanations. Consequently, the question of
whether a given kind of psychological state does, or does not, have content is (at
least to a large part) to be decided by the question of whether this kind of state
does, or does not, play a role in reasons-explanations. Reasons-explanations ‘surf ’
on the logical and inferential or evidential relations ‘induced’ between psycholog-
ical states by means of assigning contents to them.
Following this content-theoretic tradition thus suggests the following way
of looking at the question whether a given kind of psychological state has con-
tent: Do the states in question show the kind of inferential integration into their
subject’s system of propositional attitudes that would come with assigning con-
tents to them? Does the state in question provide its subjects with reasons? Is this
an important part of our folk-psychology? If the answers to these questions are
positive, we have an excellent prima facie case for assigning propositional contents
to that kind of state.
There is a complication, however: Assigning propositional content to a
kind of psychological state is necessary, but not sufficient for accounting for its
inferential integration into a system of propositional attitudes. Different kinds
of attitudes play different kinds of reason providing roles. The role of desire as
a (practical) reason provider is significantly different from that of belief. Any
account of the psychological role of a propositional attitude is thus a function
of two variables: content and attitude.
And while there are any number of contents, traditional models of theoretical
and practical reasoning, or reasons-explanation, contain only two kinds of atti-
tudinal ‘slots’: motivational and doxastic slots. So, for any propositional attitude,
it can play one of two reason providing roles: a desire-like motivational role, or a
belief-like doxastic role. To the extent that doxastic role in theoretical reasoning
amounts to a justificatory role, folk-psychology here amounts to folk-epistemology.
With these elements in place, we can now turn to our initial question: What
about perceptual experience? Does experience provide its subjects with reasons?
Content Views
80
And if so, does it fit into one of the slots provided by traditional accounts of reason
providing?
In previous writings (esp. my 2009), I have argued that it is, indeed, an in-
tegral part of our pre-theoretic conception of perceptual experience that experi-
ence provides reasons for its subject’s beliefs and actions. Like other proponents
of this claim—such as John McDowell (1994), Bill Brewer (1999), or Richard Heck
(2000)—I have focused on reasons for belief:
(R) Experience provides its subject with reasons for first-order empirical
belief.
Unlike other proponents of (R), I have combined (R) with a doxastic account of
perceptual experience, an account according to which experience is a (peculiar)
kind of belief. One of the main advantages of such an account, I have argued, is
precisely that it allows to us to understand the reason providing role of experi-
ence on the model of the reason providing role of (non-experiential) belief, thus
keeping our overall account of theoretical reasoning unified and traditional.
Availing ourselves of this advantage comes at a price, however. We only get
a plausible account of the reason-providing role of experience from a doxastic
account, if we are willing to construe experience contents in a certain non-
standard way. Standardly, experience contents are construed as being of the
‘naïve’ form x is F, where x ranges over ordinary material objects, and F over
sensible properties.6 Together with a doxastic account, this has a number of
unpalatable consequences.7 Not the least of them is that any inference from an
experience content to the most basic kind of experience based belief would be
of the ‘stuttering’ kind: It would be an inference from p to p. If nothing else, that
makes a hash of the idea that perceptual reasons are defeasible.8 Such conse-
quences can be avoided, or so I have argued, if we construe the contents of ex-
perience as ‘phenomenal contents’. Phenomenal contents ascribe ‘phenomenal
properties’ to ordinary material objects, properties such as looking F.9 In what
follows, I shall also call these contents ‘looks-contents’ or ‘Lp-contents’.
6
Strictly speaking, and on the assumption that there are sensible relations, it should be said that
experiences have contents of the form F(x1, . . ., xn), where xi (1 ≤ i ≤ n) ranges over material objects and
F over sensible properties and relations. Given the richness of experience, full experiential contents
moreover should probably be construed as (long) conjunctions of such predications.
7
For instance, it would result in outrightly contradictory beliefs in cases of known (or believed)
illusion. One might say that this is a kind of compartmentalization, a kind of compartmentalization,
moreover, that is to be expected given the ‘modularity’ of perceptual experience. Alex Byrne, who at
least seems to be quite tempted to construe experiences as beliefs, has taken this line in conversation.
8
I discuss what I call the ‘stuttering inference argument’ at some length in my (2009). The
argument derives from McDowell 1998, pp. 405f.
9
As customary, I shall focus exclusively on visual experience here. It is not completely clear how
to generalize the account to olfactory and auditory experiences as these, prima facie at least, do not
seem to take ordinary material objects.
Looks, Reasons, and Experiences 81
The doxastic account of experience I have suggested combines two elements: The
claim that experience is a kind of belief, and the claim that (visual) experience has
looks-contents:
(PB) Perceptual experience
(i) is a kind of belief
(ii) with phenomenal contents.
Of course, there are many important differences between those beliefs that are
experiences and other, non-experiential beliefs. Most importantly, perceptual
experiences have a distinctive sensory nature: Phenomenally, having an experi-
ence is very different from having a non-experiential belief.11 Experience thus is a
kind of belief, a kind that any satisfactory account of experience ultimately should
be able to specify.
10
This defense leaves any number of questions open: In what follows I do not take a stand on, for
instance, the issue of the kind of proposition best assigned as content to experience, nor the issue of
how experiential contents are represented. For all I care here, experiential representation might well
be “iconic” (Fodor 2007) or “analogue” (Dretske 1981, p. 135ff.). On my use of ‘propositional’, that
does not prevent them from having propositional content. McDowell and Crane, by contrast, recently
have come to tie having propositional content to being represented by sentence-like structures of a
language-like medium. Cf. McDowell 2008; Crane 2009.
11
This holds regardless of whether we think that there is such a thing as cognitive phenomenology.
Content Views
82
One immediate question for (PB) thus is whether (i) and (ii) are neces-
sary and sufficient for a state’s being a perceptual experience, or only necessary.
That, I think, depends on how precisely looks-contents are construed. Are there
non-experiential beliefs with looks-contents? I tend to think not. I think that the
sense in which ‘looks’ figures in the content of experiences is the so-called phe-
nomenal sense.12 In this sense, an object x cannot look F simpliciter. Rather, it
always looks F to a subject S at a time t. Moreover, x looks F to S precisely if S, at
that very time, has an experience as of x’s being F. Working along these lines, (PB)
might be able to distinguish the experiences from the rest of our beliefs by their
type of content. If that ultimately does not work, (PB) would have to fall back on
experience’s distinctive phenomenology to single out the relevant kind of belief.
Both having a distinctive type of content and having a distinctive kind of phe-
nomenology, however, are compatible with sharing the attitude component with
ordinary common and garden beliefs. Not only is this compatible, the attitude
component in perceptual experience is moreover best construed as that of holding
true or belief—or so I suggest.13 One of the most important and immediate advan-
tages of thus construing the attitude component is that it allows us to subscribe
to (R): If experiences are beliefs, they provide reasons for (further) belief in the
relatively well understood, traditional sense in which beliefs provide reasons for
(further) beliefs.14
Let me spell out in a bit more detail the sense of reason, and reason providing,
in which I take (R) to be an important and integral part of folk-psychology.15 The
12
For more on that, see below, section 4.5.
13
I am not here going to argue against any kind of sui generis account of experience that allows
experiences to be holdings true, without thereby subsuming them under the beliefs. See my (2009),
however, for an argument to the effect that the availability of doxastic accounts that preserve the special
functional role of experience undermines the very motivation for sui generis accounts.
14
The account has other advantages: It accounts for the ‘modularity’ or ‘belief-independence’
of experiences by means of their contents, not by compartmentalization, thus preserving the full
rationality of the subject of a known illusion. It allows for uniformity of contents across veridical and
non-veridical experiences. Most (virtually all) of the experience-beliefs will be true, of course, but
(PB) can account for non-veridicality in terms of misleadingness: non-veridical experiences are those
that provide their subjects with prima facie reasons for false beliefs. (PB) thus is compatible with the
intuition that non-veridicality somehow is ‘downstream’ of experience, a matter of (non-experiential)
belief (cf. Brewer 2006; Travis 2004). It also accommodates various phenomenological observations
regarding experience, for instance what is sometimes called its ‘immediacy’, its ‘presentational’
or ‘committal’ character. (PB) captures this by construing experience as belief. (PB) also captures
experience’s particularity, i.e., the claim that the veridicality of an experience depends on the intuitive
object of the experience, not on some other object that happens to make its content true. Looks-
contents naturally construe experiences as about those very objects they intuitively are about: Those
objects causally responsible for them (in the right way, of course). Moreover, (PB) accommodates
what is reasonable about transparency: Phenomenal properties, whatever their ultimate analysis, are
properties of ordinary material objects, not properties of experiences. Finally, and on the assumption
that there is such a thing as a phenomenal notion of looks, (PB) satisfies the desideratum—if it is one—
that experience content be “looks-indexed,” i.e., that experience content is determined by the looks of
things (Travis 2004).
15
All I need for my argument is that reason providing in this sense is indeed a deeply entrenched
Looks, Reasons, and Experiences 83
part of folk-psychology. I do not need to deny that there are other, more objective notions of reason,
notions that might be useful in epistemology (or elsewhere). I do not even need to deny that such
notions also are part of folk-psychology.
16
Ever since Davidson’s (1963) paper “Actions, Reasons, and Causes,” there has been widespread
consensus that reasons explanation is a species of causal explanation.
Content Views
84
here are relations between first order propositional attitudes and their contents.
A creature has reasons in the required sense as soon as its beliefs and actions can
be explained by means of its further beliefs and desires. The capacity to think
about these beliefs and desires is not required, and even less the capacity to think
about these beliefs and desires as providing the reasons in question.17 What is
required, however, is a certain minimal, subjective rationality. There are no rea-
sons explanations, be it for beliefs or actions, unless a creature’s beliefs actually to
some minimal but significant degree instantiate the basic inferential or evidential
relations objectively obtaining between the propositions that are the contents of
their beliefs.18
Now, if we think of perceptual experiences as beliefs, we can accommodate
their reason-providing role by simply extending our account of how beliefs, in
general, provide reasons for further beliefs to them. This requires some care, how-
ever, as we at the same time want to preserve certain peculiarities of the inferential
or evidential relations between experiences and other beliefs.
The two maybe most important characteristics of experiential reason pro-
viding are the following: First, the evidential relation between the contents of expe-
riences and even the most basic further beliefs based on them must be defeasible.
Experience can, and does on occasion, mislead. And when we know (or believe)
that we are in circumstances where experience is apt to be misleading, the reasons
it provides us with can be, and often are, overridden or defeated. If I know that
the light is iffy, or that there is something wrong with my eyes, I will at least be
hesitant to form beliefs about the color of nearby objects on the basis of my ex-
perience of their color. At the same time experience is such that it can, and often
does, remain completely impervious to such background beliefs about its reliability.
That’s the second characteristic feature of experiential reason providing important
17
Contrary to an anonymous referee’s objection, I do not think there is any tension between this
claim and the idea that reasons rationalize belief or action in the sense of being something that ‘speaks
for’ forming a certain belief q or performing a certain action a from the subject’s own perspective. What
we are concerned with here is first and foremost a perspective on the world, not a perspective on one’s
own mental life. What makes a reason p part of the subject S’s perspective on the world is simply the
fact that S has that reason, for instance, believes p. Then, if p is a reason for q, and S believes p, there is
something that from S’s perspective ‘speaks for’ believing q. A perspective on the world can, but does
not have to include a perspective on one’s own mental life.
That reason providing can be a purely first-order affair is not only born out by our practice of
ascribing reasons to some animals and to small children. It is also supported by the vast majority of
the reasons-explanations we give every day for the behavior of the grown-up people around us. If you
ask me why I am opening the fridge and I tell you that I am thirsty and believe there to be beer in the
fridge, my explanation is not elliptical or deficient in the sense that for it to be complete I would need to
add that I also believe that that belief and desire provide me with a reason for opening the fridge. Note,
too, that in the most basic cases an accompanying second-order belief simply cannot be required: If
believing p never rationalized believing q without also believing that believing p is a reason for believing
q, we would be off on a regress.
18
This much rationality I take to be implicit in the very content-theoretic framework employed here.
Without such rationality, reasons explanations would have no explanatory force whatsoever: Anything
could be a ‘reason’ for anything else.
Looks, Reasons, and Experiences 85
here. Take well-known illusions such as the Müller-Lyer. Those lines look as if they
were of unequal length no matter what your background beliefs tell you about their
length. And there are many very robust and stable illusions like that. Phenomena
such as these have led philosophers to think of perception as modular, or at least as
‘belief-independent’ (Evans 1982, p. 123).
None of this can be easily accounted for by a doxastic account that construes
the contents of experiences and the contents of basic experience based beliefs as
the same. I have argued that a doxastic account of experience can both preserve
defeasibility and account for the modularity phenomena if it adopts a phenom-
enal semantics for experience. Construing experiences as having looks-contents
allows these beliefs their independence from background beliefs. By the same
token, it explains why the reasons provided by such experiences can be over-
ridden or defeated precisely by considerations concerning a situation’s illusion
inducing potential. Nevertheless, or so I have suggested, experiences do provide
strong prima facie reasons for basic perceptual beliefs: That it looks as if there was
something red in front of you is a strong, but defeasible reason for believing that
there is something red in front of you.
This is where I previously left matters. In the next section, I shall consider
whether a plausible epistemology for perception-based belief can be developed on
the basis of these ideas about the reason providing role of experience.
Lp
(S)
p
19
What we want to know, that is, is when a subject S has (doxastic) justification for believing that
p on the basis of her experience—as opposed to when S is (personally) justified in forming the belief
that p on the basis of her experience.
Content Views
86
Inferences of this kind obviously will not be deductively valid. Rather, if they are
valid, they are valid in some ‘material’ sense. We can think of inferences like these
in terms of evidential support or probabilification. Such inferences are valid if the
conditional probability of p, given that Lp, is sufficiently high. They are valid, that
is, in the sense of being reliably truth-preserving.
It is at this point that the most basic worry about experiential reason providing
kicks in. It can be put in terms of a dilemma. The first horn consists of making the
inferential connection between experience and belief content ‘too tight’: If experi-
ences and basic perceptual beliefs have identical contents, the relevant inferences
‘stutter’. Defeasibility and modularity go by the board. We avoided this by assign-
ing different contents to experiences and basic perceptual beliefs. But even though
we did this by the seemingly minimal application of the looks-operator to the
content of basic perceptual beliefs, the worry now is that the gap nevertheless is
‘too wide’. The other horn of the dilemma thus consists of losing the connection
between the validity of the inference and the rationalizing power of its premises: If
nothing but brute probabilification is required, experiential premises might no
longer ‘speak for’ forming the relevant basic beliefs from the subject’s perspective
at all. Instead of stuttering, the senses now are in danger of becoming mute.20 Let’s
call this the dilemma of stuttering and silence.
We probably ought to agree that mere probabilification never is sufficient for
justification. If p and q are logically unrelated empirical propositions, believing p
just by itself never provides a subject with a good reason for believing q. But even
if this is right, it might be possible to save experiences with looks-contents from
silence.
A first, unfortunately hopeless idea would be to ‘bridge’ the gap between Lp
and p by means of further beliefs. In order to have good reason for believing the
conclusion of a ‘material’ inference, the more general reasoning here might go,
a subject not only needs to (have good reason to) believe its premise(s), but also
needs to (have good reason to) believe some principle connecting premise(s)
and conclusion. In the case of inferences following schema (S) above, the subject
could for instance believe that inferences of that form are generally reliable, or
20
Michael Pace (2008) offers an argument like this against what he calls ‘subjectivist’ accounts of
experience, i.e., sense datum and adverbial accounts. We should, he says,
capture [the] idea that experiences serve as reasons for belief. However, it is not at all clear how
subjectivist [views] can do so. The properties of which the subjectivist says one is directly aware
perceptually, whether properties of sense-data or of mental states or of ways of being appeared
to, are not the same as the properties one believes objects to have. How then can awareness of
such properties give one a reason to believe that there is something in the world instantiating the
external properties one believes to be present? (Pace 2008, p. 656)
According to Pace, the advantage of both disjunctivist and intentionalist accounts over subjectivist
ones precisely consists in construing experience as ascribing or relating the subject to the very same
properties the relevant beliefs ascribe to the objects in question (cf. Pace 2008, p. 657). His argument
might thus generalize to a ‘phenomenal intentionalism’ such as mine.
Looks, Reasons, and Experiences 87
that particular such inferences, for instance the inference from it looks as if there
is something red in front of me to there is something red in front of me, usually get
things right.21
This is hopeless, for at least two reasons. For one, it seems psychologically implau-
sible to ascribe belief in bridge principles to all subjects with good experiential reasons
for empirical beliefs. And having good reasons for believing bridge principles does not
seem any less cognitively demanding than actually believing them.22
More importantly, however, further belief clearly cannot bridge the gap in
any case. Rather, requiring belief (or having good reason for believing) in bridge
principles leads into the kind of infinite regress familiar from Lewis Carroll (1895).
Here is one way of illustrating this: As already observed, there is a (logical) gap
between the premise—Lp—and the conclusion—p—of the inference we are con-
cerned with: The inference is not necessarily truth preserving. We are trying to
close that gap by means of belief in a bridge principle. Now, assume that the bridge
principle is an inference schema like (S). Using such a schema to guide our infer-
ences can take two forms: either we treat the schema as admitting of exceptions,
or we treat it as to be followed in every case. If we treat it as admitting of excep-
tions, there will be a gap in the application of the schema to any particular instance
i: Why is i an instance where the schema will lead to a true conclusion? And if we
treat the schema as not admitting of exceptions, there is a gap in the justification of
the use of the schema itself: since it is not necessarily truth preserving why should
it be followed as if it were? In either case, there is a new gap—a gap of the very
same nature as the original gap—in need of closing by means of a further bridge
principle. And so on, ad infinitum.
We should therefore look for some other model on which to construe the
justificatory power of the ‘material’ inferences we are interested in. In what fol-
lows, I shall be solely concerned with perceptual justification. When it comes to
perceptual justification, it is very plausible to think of experiences as providing
reasons of a particular kind, it seems to me—“prima facie reasons” in the sense
21
Another idea would be that the subject has (to have good reason) to believe that experiences
in general justify believing that things are as they appear, or that particular kinds of experiences, for
instance experiences as of red things in front of one, do so. Justification derived from such bridge
principles would only be available to subjects capable of second order thought, however. This violates
the requirement that reasons, even good ones, be available to creatures without such capacity.
22
Ascribing looks-contents might seem cognitively too demanding, too, especially with respect to
the experiences of creatures of limited conceptual repertoire. Since some animals and small children are
supposed to have experiences just like ours, it is thus often argued that the contents of our experiences
must be simple enough to be plausibly ascribed to animals and small children. This argument, however,
might as well be turned around: If animals and small children indeed do have experiences just like ours
then we are fully justified to ascribe to these experiences contents precisely as complicated as required
for performing their characteristic role in us. That we thereby might ascribe some ‘powers’ to animal
experience that the animal itself does not make full cognitive use of does not seem objectionable. By
contrast, it would seem eminently objectionable to argue that animals and children do possess beliefs
in bridge principles. After all, it is quite clear that most of us do not have such beliefs, either.
Content Views
88
that are good unless defeated. Merely having these beliefs is not sufficient for
justification, but the ‘more’ that is required is not more belief, or more reason for
belief, but rather the absence of (good) reasons strong enough to defeat them. If
experiences with Lp-contents do provide Pollockian prima facie reasons, we can
avoid the stuttering of the senses without having to mute them.
But why should we think perception is special? Why should we think percep-
tual reasons are prima facie in the Pollockian sense—when so many other reasons
are not? In particular, why should we think that the absence of defeaters is suffi-
cient to turn mere probabilification into justification in the perceptual case—when
the same combination does not seem to do the trick in other cases? That this is
indeed very plausible is, I think, best brought out by considering some examples.
While many other (defeasible) reasons a subject may have for empirical belief are
not at all plausible as candidates for being prima facie reasons in the Pollockian
sense, experiential reasons are different.23
Consider Larry: Larry has never met Paul, but believes that he is left-handed.
The only ‘reason’ Larry has for this belief is his (justified) belief that Paul is red
haired. This belief clearly is not a plausible candidate for providing him with a
Pollockian prima facie reason: Intuitively, the absence of any defeaters does
nothing to make the ‘inference’ justified. Moreover, this does not change if we
assume that there in fact, but completely unbeknownst to Larry, is a strong corre-
lation between left-handedness and red hair.
Now, let’s turn the example around: Consider Laura. Laura is just like Larry
in that she does not (have any (good) reasons to) believe anything defeating the
‘hair-hand connection’, but she nevertheless never draws any conclusions about
people’s handedness from their hair color. Laura does not in any way strike us as
odd or irrational. Quite the contrary.
But the situation is rather different when it comes to experience and basic
perceptual belief. Take John. In bright daylight, he looks at a book right in front
of him. The book is red, and nothing obstructs John’s line of sight. Nor does John
believe that there is anything wrong with his eyes, or that the surrounding condi-
tions are in any way other than they seem. He does not have any reason to believe
any of this, either. Nevertheless, John does not believe that the book is red. Talking
to him about it reveals that the book does, indeed, look red to him. Asked about
defeaters, he denies believing any of them. Nor does he have any good reason
to. Yet, he assures us ardently that he does not believe the book to be red. This is
immensely odd, and quite clearly irrational.
There is a stark contrast between John and Laura: Intuitively, Laura is per-
fectly justified in not drawing conclusions about people’s being left handed from
their being red haired, while John’s refusal to draw conclusions about the book’s
23
The following discussion was prompted by conversation with Alex Byrne.
Content Views
90
(J) A perceptual experience with the content Lp provides its subject S with
justification for believing p iff
(i) Lp evidentially supports p to a sufficiently high degree, and
(ii) S does not have good reason to believe any defeaters.
24
Jim Pryor, another contemporary fan of experience as provider of Pollockian prima facie
reason, holds that it is a priori that experiences as of p in the absence of defeaters justify believing that
p (cf. Pryor 2000).
25
No experiential prima facie reason is such that it is a good reason for believing something
simply by virtue of the contents believed. It is good only in the absence of defeaters. Defeaters of an
eminently empirical nature, moreover.
Looks, Reasons, and Experiences 91
The presence of a relation of evidential support (of sufficiently high degree), how-
ever, depends on the way the world actually is. If the world ‘cooperates’, schema
(S) inferences will be reliable. It is in this sense, and in this sense only, that schema
(S) reasoning is justified or warranted: In the absence of defeaters, such reasoning
is warranted by its reliability. This much ‘externalism’, it seems to me, is ultimately
unavoidable in the theory of empirical justification or warrant. Combining the
phenomenal belief account with this account of experiential justification, we can
avoid the dilemma of stuttering and silence. If the senses provide us with defea-
sible prima facie reasons, they neither stutter, nor are they mute. Whether what
they ‘say’ can be trusted ultimately remains hostage to the world. But that is how it
should be, it seems to me—it’s the human predicament.26
In this final section, I want to take up another challenge to the phenomenal belief
account. So far, my overall claim has been that to the extent that we can provide a
plausible account of the intuitive inferential integration of experience into systems
of propositional attitudes, we thereby provide the strongest motivation available
for construing experience as having content in the first place. I have argued that
the phenomenal belief account does precisely that: By construing experiences as
beliefs with phenomenal contents it accounts for their intuitive inferential inte-
gration without either jeopardizing modularity or falling prey to the dilemma of
stuttering and silence. Silence is avoided by understanding experiential reasons as
26
It is fairly obvious that there is no (direct) anti-skeptical mileage to be gotten from the idea that
experience provides reliable prima facie reasons for (further) belief. This becomes drastically clear once
we spell things out in terms of probabilities. Plausibly, reason (or evidence) providing is governed by
the following principle (cf. Carnap 1950, pp. 382ff.; Spectre 2009, pp. 91ff.):
(EP) r is a reason for s only if Pr(s/r) ≥ Pr(s).
But now consider the following example:
(Lp) It looks as if there is something red in front of you.
(p) There is something red in front of you.
(q) There is something white in front of you that is illuminated with red light.
It clearly holds that Pr(p/Lp) ≥ Pr(p). But it holds equally clearly that Pr(q/Lp) ≥ Pr(q). Moreover,
Pr(q/Lp) clearly is greater than Pr(q), which means that it is not the case that Pr(¬q/Lp) ≥ Pr(¬q).
Consequently, that it looks as if there is something red in front of you if anything provides you with a
reason for, not against believing that there is something white in front of you that is illuminated with
red light. Of course, this reason will (normally) be much weaker than that simultaneously provided
for believing that there is something red in front of you, but nevertheless: Experiences do not provide
reasons against phenomenally compatible skeptical hypotheses.
Davidsonians might think that—despite the ‘contingent’ nature of schema (S) inferences—
general considerations of content determination might help. While I have my doubts about that, the
phenomenal belief account of experience is perfectly compatible with both a Davidsonian account of
content determination and Davidsonian epistemology in general. For more on this, see my (2012) and
Stroud (2002).
Content Views
92
prima facie reasons in the Pollockian sense. A phenomenal semantics for experi-
ence thus does not only not obstruct a plausible epistemology of experience-based
belief—it makes providing one quite easy.
The challenge I want to consider now is the following: Even if phenom-
enal contents allow us to get the epistemology of experience-based belief right,
experience cannot plausibly be construed as having phenomenal contents. The
looks-operator used by the phenomenal belief account is modeled on the so-called
“phenomenal” use of looks-locutions in natural language, specifically English. And
in English, phenomenal ‘looks’ is itself a propositional attitude operator. An oper-
ator modeled on English phenomenal ‘looks’ therefore cannot be used to specify
the content of any first order propositional attitude.27 Let’s call this the ‘attitude
operator argument’.
I shall argue that phenomenal ‘looks’—‘looksp’—is not a propositional atti-
tude operator. The argument will be simple in structure: Propositional attitude
operators create hyperintensional contexts. But ‘looksp’ does not. Therefore, it is
not a propositional attitude operator.
I take it that the assumption that any propositional attitude operator worth its
name creates hyperintensional contexts is uncontroversial. What is controversial
is the claim that ‘looksp’ does not. In fact, observations to the effect that perception
verbs do create such contexts have been used in the literature to defend the very
claim that perceptual experiences have contents at all, and also that they have con-
tents of a particular kind. In Searle, for instance, we find the following argument:
An additional clue that the ‘sees that’ form expresses the Intentional content
of the visual experience is that this form is intensional-with-an-s . . . . The most
obvious explanation of this . . . is that the ‘sees that’ form reports the Intentional
content of the perception. (Searle 1983, pp. 41f.)
This, however, is not the best of arguments. For one thing, intensionality is not
sufficient; (alethic) modal operators such as ‘it is necessary that’ create intensional
contexts, and in their case, the most obvious explanation is not that the ‘it is nec-
essary that’ form reports the content of any mental state. In order to have a better
argument, we should require the creation of hyperintensional contexts. But even if
‘sees that’ creates such contexts, we still would not have a good argument for the
claim that this form reports the content of perception. Arguably, ‘sees that’ implies
belief, and if it does, ‘sees that’ reports probably report not the content of experi-
ence but the content of beliefs formed on the basis of experience.
Nevertheless, a better argument might seem to be very close by: All we need to
do is replace ‘sees that’ by ‘it looksp as if ’. For those doubting the hyperintensionality
27
Arguments like this have been suggested to me by several people in conversation, first by
Susanna Siegel.
Looks, Reasons, and Experiences 93
created, examples of substitution failure of the relevant kind seem readily avail-
able. For instance:
(1) It looks to Lana Lang as if Superman is flying by.
(2) It looks to Lana Lang as if Clark Kent is flying by.
That (1) is true in no way guarantees that (2) is. Assuming that ‘it looks as if ’ in
sentences such as (1) and (2) is used phenomenally and as a propositional attitude
operator, Brogaard has argued that the content of visual experience, just like the
content of belief, is Fregean (cf. Brogaard 2011a). While I agree with a slightly more
careful claim—that experience content, just like (all other) belief content, cannot
be modeled by possible worlds propositions alone—I do not think that ‘looksp’
creates propositional attitude contexts. Examples like (1) and (2) are misleading.
Before we look into that, however, a few words characterizing the phenomenal use
of ‘looks’ are in order.
Following Chisholm (1957) and Jackson (1977), it has become customary to
distinguish between at least the following three uses of phrases like ‘it looks as if
p’, ‘x looks F’, or ‘x looks like an F’: The ‘epistemic’ use, the ‘comparative’, and the
‘non-comparative’ or ‘phenomenal’ use. While none of these uses seems to have
any dedicated grammatical form (cf. Brogaard 2011b), they can be illustrated by
means of example. For instance,
(3) It looks as if the neighbors are away,
is most naturally used epistemically. Using (3) epistemically, the speaker says
something like that she has good reasons to believe that the neighbors are away.
These reasons can, but do not have to be visual, or even perceptual—as illus-
trated by saying ‘it looks as if Obama won the election’ after listening to the
news on the radio. On the other hand, (4) is most naturally used comparatively:
(4) The neighbor’s car looks like a tank.
Comparative uses arguably are best analyzed as existentially quantifying over
ways of looking, or over looks: Roughly, there is a way of looking such that both
the neighbor’s car and tanks have it (Byrne 2009; Brogaard 2011b). That is, the
car and the tank are compared with respect to their looks, and found to be alike.
What exactly the similarity consists in is, of course, a matter of context; it is often
suggested that the object of comparison (normally) is the way a certain kind of
object, here tanks, normally or under standard conditions look (to normal sub-
jects). But we can easily think of contexts in which the object of comparison is the
way these things look under non-standard conditions. If you are looking for red
apples on a dark summer evening, for instance, you might well use ‘that looks like
a red one’ to say that the demonstrated apple has the look that red apples on dark
summer evenings have (cf. Jackson 1977, p. 32; Chisholm 1957, p. 46).
Analyzing the comparative use of ‘looks’ along these lines, we make use of
‘looks’ in the analysans, however. To complete the analysis, we need an explanation
Content Views
94
of what these ways of looking are, an explanation that does not itself make (com-
parative) use of ‘looks’. It is therefore often argued that the comparative use of
‘looks’ presupposes another, third use of ‘looks’. This third use is then identified
as the non-comparative, or phenomenal use (cf. Maund 1986, p. 171; Byrne 2009,
p. 441; Brogaard 2011b). This amounts to analyzing comparative looking in terms
of comparisons between non-comparative or phenomenal looks.
The crucial question for any completion of the analysis of comparative
‘looks’ is the following: What is it that comparative uses of ‘looks’ quantify
over? What are these ways of looking that get compared when we compare the
looks of things? Intuitively, the look of an object o is a property of o that varies
with the conditions under which o is viewed. Thus, a white object can look red
if viewed in red light. And a red apple can look a certain shade of grey when
viewed on a dark summer evening. Moreover, such looks intuitively also vary
with the viewer; things look different with my glasses on or off, and the way a
red apple looks to my color blind father might well be the way it looks to me on
dark summer evenings. A very natural idea therefore is to ‘phenomenalize’ looks
along the following lines: Ways of looking are relational properties of material
objects, properties somehow involving the experiences of subjects looking at
these objects. They might, for instance, be dispositions to cause experiences of
certain phenomenal kinds under certain conditions. Or they might be properties
objects have precisely when causing experiences of certain phenomenal kinds.28
Whatever looks precisely are, the basic idea behind the claim that the compar-
ative use presupposes the phenomenal use of ‘looks’ is that what gets compared by
comparative uses are precisely such phenomenalized looks. But when using ‘looks’
comparatively, the way the object looks remains (literally) unspecified: What is
28
Martin (2010) also suggests that comparative ‘looks’ be analyzed as existentially quantifying over
looks. But according to him, looks are intrinsic properties of objects, more precisely, they are identical
to the basic visible properties such as redness and squareness (Martin 2010, pp. 161, 207ff.). Moreover,
Martin argues, most ‘looks’-statements—even those not obviously comparative on the surface such as
‘o looks red’—are to be analyzed as comparative, i.e., as comparing an object’s look to that characteristic
of a contextually relevant class of objects. According to Martin, there is no semantic reason to prefer
an account according to which looks are relational properties involving the phenomenal character
of our visual experiences of objects to what he calls “parsimony,” i.e., to identifying looks with basic
visible properties (cf. Martin 2010, p. 222). And indeed, the semantics he suggests can account for
our intuition that ‘o looks bent’, said of a straight stick halfway immersed in water, is true. According
to Martin, we here do compare the stick’s straightness to the characteristic look of bent things, i.e.,
bentness, but we do so with respect to a contextually fixed similarity measure: “The stick is similar to
bent things simply with respect to how it strikes me, or the subjective bearing it has on me” (Martin
2010, p. 215). This requires the semantics to be doubly context-dependent, however: Martin needs
both a contextually fixed comparison class, and a contextually fixed similarity measure to get the
intuitive truth values of ‘looks’-sentences right. In my (2013) I argue that there therefore is semantic
reason to prefer construing looks as relational properties involving the phenomenal character of our
experiences—because it allows for a significantly simpler semantics for ‘looks’-sentences. I also argue
that, contrary to appearances, “parsimony” has no advantage when it comes to explaining the intuition
(if it is one) that things have looks even if no-one is around to see them (cf. Martin 2010, pp. 209, 220).
Looks, Reasons, and Experiences 95
‘said’ is that there is a way of looking that the object shares with (an)other object(s)
with a certain property (cf. Maund 1986; Byrne 2009, p. 440). In phenomenal uses
of ‘looks’, by contrast, the way an object looks gets specified, or referred to, in a
(more) direct way.29
It seems very plausible to me that we make phenomenal use of ‘looks’ in nat-
ural language. There are some difficult questions regarding our use of ‘looks’, ques-
tions like whether the phenomenon we are observing is one of genuine ambiguity
or polysemy, and how to secure a compositional semantics for phenomenal ‘looks
F’.30 For the purposes of this chapter, we can work with the understanding of phe-
nomenal ‘looks’ developed so far. I’ll assume that there is such a use in natural
language and I shall indicate it by means of ‘looksp’.
I should point out, however, that for me, nothing really hangs on this latter
claim. Should it turn out that there is no phenomenal use in natural language, I’ll
just define a phenomenal looks operator (and predicate modifier) that works in the
way indicated and use it in my semantics for experience.31 This might (or might not)
pre-empt the attitude-operator argument, but if it does, so much the better for me.
Jackson thought that the phenomenal use of ‘looks’ is restricted to, and maybe
even induced by, combining ‘looks’ with predicates of color, shape, and distance
(1977, p. 33). However, it seems plausible that sentences of the form
(5) x looks red,
can be used comparatively (and probably even epistemically). It also is very plau-
sible to think that sentences of the form
(6) x looks old,
can be used non-comparatively or phenomenally. Consider (7):
(7) x looks red and old.
29
For those of us impressed by the possibility of inverted spectra (or inverted phenomenal
qualities more generally), there is a complication here: There is no ‘direct’, intersubjectively accessible
way of specifying phenomenal kinds of experiences. We can still refer to any phenomenal kind,
however, for instance by specifying it functionally as that phenomenal kind that in the subject plays
a certain epistemic or reason-providing role. For a construal of sensation terms that could be used
as a model here, see Pagin 2000. In recent conference talks, both Maund and Pagin have suggested
analyzing phenomenal ‘looks’ along such lines.
30
Byrne (2009, p. 444) suggests in passing that ‘looksp F’ is “idiomatic in the interesting way ‘red
hair’ is.” Given the great variability of ‘F’ here, that would be bad news for the compositionality of
natural language. As hinted in the previous note, it might be possible, however, to analyze phenomenal
looks in terms of prima facie reason providing. Such an analysis might preserve the compositionality
of ‘looksp F’. My own preferred solution acknowledges that ‘looksp F’ is not compositional (in the
traditional sense). But it is not idiomatic, either. Instead, it is general compositional—which arguably
is just as good (see below, n. 39 for a tiny bit more on this. See Pagin and Westerståhl 2010a, b for the
notion of general compositionality). This is a topic for another paper, however.
31
Of course, the precise semantics for this operator needs to be worked out more precisely. But
that holds whether or not this operator exists in natural language.
Content Views
96
(7) is clearly well-formed, and that means that there is at least one uniform inter-
pretation of (7). To me, it seems more difficult to read (7) as uniformly compara-
tive than to read it as uniformly phenomenal. If read phenomenally, what we say
by means of (7) is that x both has the ‘red-look’ and the ‘old-look’ (cf. Byrne 2009,
p. 442; example originally from Thau 2002, p. 230).
Byrne (2009) uses observations like these to make trouble for the idea
that phenomenal looks “index” the content of experience.32 We cannot read off
experience contents from phenomenal uses of ‘looks’, Byrne argues, because
objects can lookp F without being represented as F by the subject’s experience.
Take Byrne’s own example: naked mole rats. Naked mole rats are bald, pink,
and wrinkled; they lookp old no matter how old they are. All their life, that is,
naked mole rats have the ‘old-look’. But, Byrne submits, that does not mean
that experience represents them as old. Rather, experience represents naked
mole rats as bald, pink, and wrinkled. Using “exing” as his term for the expe-
riential propositional attitude, Byrne concludes: “If a naked mole rat looksnc
old to S, then S exes, of the rat, that is wrinkled, pink, etc.—not that it is old”
(p. 443).33
Even though Byrne thinks that the possibility of using ‘looks old’ phenom-
enally thus spells trouble for the idea that ‘looksp’ “indexes” the content of visual
experience, he probably does not think that these observations ultimately pre-
vent us from construing ‘looksp’ as a propositional attitude operator: Whenever
something looksp F to a subject S, there is a p such that S exes that p. It’s just
that p does not have to be x is F. Nevertheless, I think Byrne’s observations
do provide us with some important clues here. They point us towards features
of ‘looksp’ that ultimately undermine its construal as a propositional attitude
operator.
Take some naked mole rat, for example, and call it Mora. By means of (8)
(8) Mora looksp old,
I can ascribe the old-look to Mora. But I could ascribe the very same look to Mora
by means of (9):
(9) Mora looksp bald, pink, and wrinkled.
32
Travis (2004) argues that there is no notion of looking that “indexes experience content,”
i.e., determines a unique content for a given experience. He does not consider phenomenal looking,
however, and several authors have claimed that phenomenal looks do “index” experience content (cf.
Brogaard 2011b; Schellenberg 2011). Byrne takes issue with this claim; according to him, not even
phenomenal looks “index” experience contents.
33
As already noted above (n. 12), the phenomenal belief account of experience gets around the
problem of indexing (if it is one). Phenomenal contents are indexed by phenomenal ‘looks’—rather
trivially so. Phenomenally speaking, ‘looks old’ and ‘looks bald, pink, and wrinkled’ are equivalent;
both are ways of specifying the very same phenomenal look, and, thus, the very same experience
content—according to the phenomenal belief account. This is precisely the feature of ‘looksp’ that we
shall use in what follows.
Looks, Reasons, and Experiences 97
34
There is, of course, more than one way of lookingp old. Houses, for instance, lookp old in a
different way than mole rats do. Which way of lookingp old is the relevant, or salient, one depends on
the context. I shall abstract from this complication here.
35
On the assumption that a subject S needs to know what lookingp F is like (to S) to fully
understand the predicate ‘looksp F’, it seems reasonable to assume that if F and G are indeed
phenomenally equivalent, no rational subject understanding both ‘looksp F’ and ‘looksp G’ will assign
different truth values to ‘o looksp F’ and ‘o looksp G’.
36
It is an interesting question which predicates are what we might call “phenomenal,” i.e., are such
that there are, or can be, objects satisfying the complex predicate formed by means of combining them
with the predicate modifier ‘looksp’.
Content Views
98
to Gretria. But as shown by the example we started with, the example of Mora,
predicates do not have to be necessarily co-extensional for there to be contexts in
which they are phenomenally equivalent. Moreover, they do not even have to be
co-extensional. But if two predicates are phenomenally equivalent, they can be
substituted salva veritate in looksp-contexts.
Phenomenal equivalence is not restricted to predicates. Proper names can
be co-phenomenal, too. Take two qualitatively identical tomatoes, Tim and Tom.
Their names, like many others, can be used to ascribe looks to objects. Tim and
Tom are interesting, however, because the Tim-lookp clearly is the same as the
Tom-lookp. ‘Tim’ and ‘Tom’ are co-phenomenal proper names. But Tim and Tom
are different tomatoes. ‘Tim’ and ‘Tom’ do not co-refer. Consequently, ‘Tim’ and
‘Tom’ are not co-intensional, either. Nevertheless, (12) is true whenever (13) is. In
looksp-contexts, ‘Tim’ and ‘Tom’ are intersubstitutable salva veritate:
(12) It looksp as if Tim is sitting on the table in front of me.
(13) It looksp as if Tom is sitting on the table in front of me.
This is in stark contrast to (1) and (2), of course. ‘Superman’ and ‘Clark Kent’ do
co-refer, but cannot be substituted salva veritate in (1). And on the assumption that
‘Superman’ and ‘Clark Kent’ have the same intension—an assumption that I do not
share, but will not challenge here—the contrast is even starker. But now, we can explain
why ‘Superman’ and ‘Clark Kent’ cannot be substituted salva veritate in sentences
such as (1)—without having to say anything about (1) being a propositional attitude
context: ‘Superman’ and ‘Clark Kent’ cannot be substituted salva veritate in (1) be-
cause the Superman-lookp is very different from the Clark Kent-lookp: ‘Superman’
and ‘Clark Kent’, even though co-referential and co-intensional, are not only not
co-phenomenal—they are such that nothing can have both looks at the same time.37
Co-phenomenality thus is rather special when compared to co-extensionality.
More precisely, co-phenomenality and co-extensionality appear to be independent
properties. Proper names can be co-extensional and co-phenomenal, but they
can also be co-extensional without being co-phenomenal, and most importantly,
they can be co-phenomenal without even being co-extensional. The same holds
for predicates. For (phenomenal) predicates, it also seems to hold that they are
co-phenomenal if they are necessarily co-extensional. These observations suffice
for present purposes. The claim I have put forward is the following:
(LP) Co-phenomenal expressions can be substituted salva veritate in
looksp-contexts.
37
Even though Superman both can have the Superman-lookp and the Clark Kent-lookp, he cannot
have them at the same time. So, even though it is true that Superman (sometimes) looksp like Clark
Kent (and vice versa), substitution of the names within looksp-contexts not only can result in truth value
change—it always does. If it is true that Superman looksp like Clark Kent at time t, it is not true that he
looksp like Superman at t. And vice versa. Note, however, that there are plenty of non-identical looksp
that an object can have at the same time, for instance the red-lookp and the old-lookp. Substituting ‘red’
for ‘old’ in a looksp-context thus does not necessarily change truth value.
Looks, Reasons, and Experiences 99
38
In her Introduction to this volume, Berit Brogaard points out that there is an understanding of
hyperintensionality on which an operator is hyperintensional “just in case substituting an expression
for a logically (or metaphysically) equivalent expression under the operator changes the truth-value
of the whole” (p. 11). Let’s call any understanding of hyperintensionality according to which a
context is hyperintensional iff substitution of “co-intensional” expressions (in the relevant sense of
‘co-intensional’) can result in truth value change “weak hyperintensionality.” On the assumption
that ‘Superman’ and ‘Clark Kent’ are indeed co-intensional (in the relevant sense), ‘looksp’ is weakly
hyperintensional simply because substituting the one for the other in sentence (1) does result in truth
value change.
So, what did I have in mind when arguing that ‘looksp’ is not hyperintensional? Well, first of all,
I was thinking of extensional, intensional, and hyperintensional contexts along the following lines.
In extensional contexts, co-extensionality suffices for salva veritate substitutability. In intensional
contexts, it does not. What suffices instead, is co-intensionality (the relevant form of which, of course,
can be understood in different ways—e.g., as sameness of classical possible worlds intension or, on
two-dimensionalism, as sameness of secondary intension or, on evaluation switcher semantics, as
sameness of actualist intension). But co-intensionality entails co-extensionality. And analogously
for hyperintensional contexts: Here, co-hyperintensionality (whatever that amounts to) suffices for
substitutability. But co-hyperintensionality entails co-intensionality and co-extensionality.
Thought of along these lines, hyperintensionality requires more than weak hyperintensionality: it
also requires that what suffices for substitutability entails co-intensionality and co-extensionality.
Let’s call this “strong hyperintensionality.” My point then is that, while weakly hyperintensional,
looksp-contexts do not seem to be strongly hyperintensional: Looksp-contexts are such that what suffices
for substitutability is co-phenomenality. And co-phenomenality does not entail either co-intensionality
or co-extensionality. In looksp-contexts, ‘Tim’ and ‘Tom’ are substitutable, and so are ‘old’ and ‘bald,
pink, and wrinkled’ (in the context of describing Mora’s looks).
Secondly, it is precisely with respect to strong hyperintensionality that looksp-contexts
differ from paradigmatic propositional attitude contexts such as belief contexts. These are strongly
hyperintensional. Moreover, strong hyperintensionality would seem to be required for being a
propositional attitude operator. Looksp-contexts thus share a certain more superficial characteristic—
weak hyperintensionality—with propositional attitude contexts, but deeper down, it seems to
me, they are a very different kind of animal. So, whatever we ultimately think ought to be called
‘hyperintensionality’ (and Brogaard may well be right that her understanding is more standard than
what I had in mind), an operator that does not create strongly hyperintensional contexts is not a
propositional attitude operator.
Why would being a propositional attitude operator require strong hyperintensionality?
Roughly, because the use of expressions within the scope of propositional attitude operators can be
sensitive to the precise content of the attitudes ascribed by means of them. It is a consequence of such
Content Views
100
In this chapter, I have further developed and defended the phenomenal be-
lief account of perceptual experience I have suggested earlier. In particular, I
have defended the account against two objections against phenomenal contents:
The objection that phenomenal contents prevent us from developing a plausible
epistemology for perception based belief, and the objection that phenomenal
‘looks’ cannot go into the content of experience because it is a propositional
attitude operator. Construing experiences as beliefs with phenomenal contents
allows us to account for the intuitive inferential integration of perceptual ex-
perience into our systems of beliefs and other propositional attitudes. This in-
ferential integration in turn provides us with one of the best motivations for
construing experience as having propositional contents in the first place. By of-
fering an account of this inferential integration I have thus ipso facto defended
the claim that experience indeed has content.
References
“content-sensitivity” that substitutions within the scope of an attitude operator that do not preserve
ascribed content can change the truth value of the ascription. And ascribed content is preserved across
substitution only if substituted and original expressions are both co-intensional and co-extensional.
It is precisely this kind of content-sensitivity that expressions within the scope of ‘looksp’ do not
exhibit. As pointed out, in looksp-contexts, co-phenomenality suffices for salva veritate substitutability,
but co-phenomenality does not require either co-intensionality or co-extensionality. It is this rather
peculiar behavior of the looksp-operator that I wanted to draw attention to. This behavior, I think,
precludes the looksp-operator from being a propositional attitude operator.
But why would ‘looksp’ behave in this peculiar way? My hypothesis right now is that ‘looksp’
is such that co-phenomenal expressions within its scope have the same meaning. For instance,
if F and G are co-phenomenal predicates, the expressions ‘looksp F’ and ‘looksp G’ mean the same.
Precisely how that works is a subject for another paper. Pace Byrne (cf. above, n. 31), I think we
should prefer the meaning of the complex looksp-expressions not to be a matter of idiomatic usage. To
achieve the desired identities of meaning, however, we cannot construe the meaning of the complex
looksp-expressions compositionally. Instead, we need to think of ‘looksp’ as what Peter Pagin and I have
called an “evaluation switcher”: an operator that switches the function used to semantically evaluate
what is in its scope. The resulting semantics won’t be compositional (in the traditional sense), but it
will be what Pagin and Westerståhl call “general compositional”—which arguably is just as good (for
more on switcher semantics and general compositionality, see Glüer and Pagin 2006, 2012; and Pagin
and Westerståhl 2010a, b, c).
Looks, Reasons, and Experiences 101
Pagin, Peter, & Westerståhl, Dag (2010b). Compositionality II: Arguments and problems.
Philosophy Compass, 5, 250–264.
Pagin, Peter, & Westerståhl, Dag (2010c). Pure quotation and general compositionality.
Linguistics and Philosophy, 33, 381–415.
Pollock, John (1974). Knowledge and Justification. Princeton: Princeton University Press.
Pryor, J. (2000). The skeptic and the dogmatist. Nous, 34, 517–549.
Schellenberg, Susanna (2011). Perceptual content defended. Nous, 45, 714–750.
Spectre, Levi (2009). Knowledge Closure and Knowledge Openness: A Study of Epistemic
Closure Principles. PhD thesis. Stockholm University.
Stroud, Barry (2002). Radical interpretation and philosophical scepticism. In id.,
Understanding Human Knowledge: Philosophical Essays (pp. 177–202). Oxford: Oxford
University Press.
Thau, Michael (2002). Consciousness and Cognition. New York: Oxford University Press.
Travis, Charles (2004). The silence of the senses. Mind, 113, 57–94.
PART TWO
The main argument of this paper was presented in a symposium on whether perception is
predicative, held at NYU in 2010. I thank Adam Pautz for his comments on that occasion, as well as
the other discussants at NYU, particularly Ned Block who pressed me on the issue of unconscious
perception, and Tyler Burge who emphasized the distinction between the individuation conditions of
an experience and its essential conditions of realization. 105
Against Strong Content
106
of our whole practice of thinking and talking about experience and its epistemic
significance?
In order to begin to address that pivotal question, surely one of the most
important in the philosophy of perception, we need to examine (i) just how we
frame such pre-judgmental experiential conditions in ordinary language when
we report our sensory episodes and (ii) how we appeal to such conditions, so
framed, in our folk epistemology. Only then can we enquire after the ontological
status of such conditions or states or events in the light of the details of the psy-
chology of perception. Only then will it be worth pursuing questions of emer-
gence, reduction and potential elimination of the items ostensibly recognized in
our ordinary thought and talk about experience, and most importantly in our
ordinary attempts to articulate our entitlement to believe this or that on the basis
of what we have experienced.
When we report the sensory experiences of ourselves and others, we often
resort to object-directed idioms in order to characterize datable sensory episodes.
For example:
Paul gazed at the Pantheon in amazement until the tour guide interrupted
him.
Jane briefly smelled the coffee, and then took a cup.
For two minutes, Uri watched the Rottweiler chewing the meat.
Sam listened to Sutherland’s vocal acrobatics, until he could stand them no
more.
Mary tasted the astringency of the calvados and then spat the drink out.
Suddenly, Fred’s attention was captured by the brightness of the moon’s reflec-
tion in the water.
We also report subjects perceiving things as thus and so.
Paul saw the man as waving for help.
Mary felt the texture of the curtain as rough.
Fred heard the key of the piano trio as E-minor.
Then there are propositional attitude reports of experience, for example:
Paul saw that the man was waving for help.
Mary felt that the texture the curtain was rough.
Fred heard that the key of the piano trio was E-minor.
But these last reports are curiously promiscuous. As has been widely observed
from Roderick Chisholm (1957) on, the very same propositional attitude reports
can be used to describe a variety of non-sensory ways of finding out that things
are thus and so, be it by insight, inference or by testimony. You can hear that the
president was shot, without hearing the shooting. You can see that the economy
The Problem with the Content View 107
The now-standard content view in the theory of perception, a view that counts
a certain type of propositional attitude as central to experience, is most clearly
set out by Alex Byrne (2009) in his characteristically incisive “Experience and
Content”. One of Byrne’s ambitions there is to separate the standard view from
any naïve commitment to introspectable experiential episodes. Byrne is prepared
to doubt that there are such things, and yet still set out and argue for the standard
view of experience.
We happen to disagree on that point, since I believe that experience is best
understood in terms of attentive sensory episodes, and these are introspectable,
datable events—for example, on the basis of introspection, I can give an account
of roughly how long I visually attended to the butterfly’s sitting on my leg—but this
is not the main issue in what follows.
To explicate the standard content view, Byrne proposes that perception consti-
tutively involves a non-factive propositional attitude rather like the non-factive atti-
tude of believing: he calls it EX-ing, meant to suggest experiencing. This non-factive
propositional attitude is directed at a propositional content, which is true when the
perception in question is veridical and false when it is non-veridical. The content of
my EX-ing at any given time is a proposition which states how the scenarios before
my senses would have to be if the EX-ing in question is to be veridical. EX-ing is thus
an attitude directed at its own success conditions, at least in so far as those conditions
are conditions of veridicality. This is why EX-ing has to be non-factive, since the con-
tent EX-ed can be true or false depending on whether the EX-ing is veridical or not.
Indeed, on this model of perception, the veridicality of experience just is the truth of
the proposition EX-ed. Byrne adds:
One may think of the content of the EX-ing attitude as the output of (largely)
informationally encapsulated perceptual modules. Sometimes one will be in
possession of background information that undermines that q; that will not
affect the output, resulting in the subject EX-ing that q while disbelieving it.
CV, as just explained, is intended as a theoretically fruitful description of
the phenomenon of perception, not a piece of unarticulated folk psychology . . .
Various optional extras can be added as desired: that the relevant con-
tents are “non-conceptual”, that there’s a different attitude for each of the dif-
ferent perceptual modalities, and so on. For present purposes, though, we can
work with CV in skeletal form.
Finally, it should be emphasized that the exposition of CV is here is
entirely unoriginal, and merely repeats with minor amendments a charac-
terization that is often found in the literature. For instance, in On Clear
and Confused Ideas (p. 111), Millikan [2000] introduces “visaging”, “a gen-
eral term for what stands to perceiving as believing stands to knowing”; to
suffer a perceptual illusion is to “visage falsely”. And Johnston [1997], in a
Against Strong Content
110
postscript to his paper ‘How to Speak of the Colors’ (pp. 172–173), discusses
the view that visual experience involves “a sui generis propositional atti-
tude—visually entertaining a content concerning the scene before the eyes.”
As it happens, I did take the sui generis character of the putative attitude to raise
some suspicions about its existence, but Byrne allays this concern by insisting
that we should construe EX-ing as a theoretical posit in an account of just what
experience consists in. The question arises: is this a theoretical posit in percep-
tual psychology, in which case what is the detailed empirical evidence for it? Or
is it a theoretical posit that helps us to regiment our folk psychological perceptual
notions, in which case why is it useful as such? I take it to be the latter, and I take
Byrne and many others to be describing just why it is useful as such.
Here is one way of getting a feel for the content view. As I walk down the
street the scene changes, indeed there is continuous alteration in all of the sce-
narios before my senses. Some of these changes I perceptually register and these
determine the character of my whole course of experience. Think of that whole
course of experience as I walk down the street as modeled by a series of EX-ings
on my part directed at different contents in accord with the changes that I am per-
ceptually registering. As Byrne sees it, there may be no discrete or distinguished
introspectable sub-events within a course of experience, events which themselves
count as experiences in any important theoretical sense.
Byrne then goes on to offer an interesting account of the kinds of contents
that can figure as the propositions EX-ed. Following Roderick Chisholm (1957)
he distinguishes:
(i) Epistemically grounded uses of “looks” as in “It looks like our suspect is
Scandinavian, since his name is Sven Engstrom.”
(ii) Comparative uses of “looks” in which the speakers intent is to compare
the visual appearance of something to a typical appearance of a kind of
thing, as in “Oddly, in this light that Irishman looks like a Scandinavian.”
(iii) Non-comparative (and non-epistemic) uses of “looks”, such as “He
looks Scandinavian”; where this involves the speaker focusing on a
distinctive kind of visual gestalt made up of “mid-level perceptual
features” such as shape, size, motion, color, shading, texture,
orientation, timbre, loudness, pitch and the like, the very kinds of
features perceptual illusions prompt us to misattribute to objects.
Many ordinary uses of “looks red”, “looks round”, “feels soft” etc. are
non-comparative uses. But there is also such a use of “looks old” and
“looks expensive”, at least if old things and expensive things have
characteristic “looks” or visual gestalts.
Byrne argues against Susanna Siegel (2006) and others that the content of EX-ing
and hence of perception properly speaking is fairly minimal, in that the content
of EX-ings are confined to predications involving the features that can figure in
The Problem with the Content View 111
over there is visually experiencing that the cat is over there and (Searle’s self-referential
addition) that the cat is causally responsible for the very experiencing in question.
Searle’s argument that experience has success conditions, and that we should
therefore model specific experiences as directed at their success conditions is now
widely recognized as resulting in contents too recherché for any genuinely sensory
act to have. Searle himself notes that one way an experience might fail is for it to be
a veridical hallucination: you might hallucinate a cat before you, and by accident
there might be a cat before you. Hence his thought that the content of the relevant
experience should not count this experience as a success, and so the content of the
experience should include a self-referential causal condition to the effect that there
is a cat before one causing the very experience in question. There is some plausi-
bility in the idea that perceptual content might include some idea of dependence
of experience on the external world, even perhaps a kind of causal dependence of
the sort Searle emphasizes.
Yet clearly, the self-referential causal condition is not enough to capture the suc-
cess conditions of experience; in particular it is not enough to count all cases of
hallucination as perceptual failures. There is a quite restrictive range of causal condi-
tions required to rule out hallucination, as is shown by the fact that a hallucination
of a cat before us can be caused by a cat before us. This is the old and quite general
problem of “wayward causal chains”. As will emerge later, what is implausible is that
any condition at odds with every wayward causal route to an experience is part of the
content of that experience. Accordingly, the success conditions of an experience are
not captured by the propositional content associated with that experience.
A second unconvincing argument offered by Searle for the content view is
that experience can be less than fully determinate in various ways. You might see a
speckled hen, and see it as having a lot of speckles without seeing it as having some
number n of speckles. But macroscopic reality is here fully determinate: the hen
itself has some number n of speckles. How does the merely determinable element
get into perception? The friend of the content view says that the things predicated
in the propositions that give the content of experience can be more or less deter-
minate conditions.
As is now widely recognized, this argument from the merely determinable char-
acter of experience is at most an argument for the conclusion that experience is inten-
tional in the sense of presenting items in the external environment in certain ways,
or under certain modes of presentation, where such ways or modes can be more or
less determinate. The argument does not decide between a propositional model of
intentional directness towards a not-fully- determinate content and a presentational
model of directedness to an object—the hen—under a not-fully-determinate mode
of presentation, namely as having a lot of speckles (Crane 2006, 2011).
Another influential but unsuccessful argument for the content view moves
from the fact that experiences can be illusory or non-illusory, to the claim that this
“bipolarity” of experience is best modeled by taking experience to constitutively
involve EX-ings that p.
The Problem with the Content View 113
by an objective array which does satisfy the manner of presentation typical of the
standard illusion. Here the experience is a veridical illusion, i.e. there is something
in the scene before the eyes which satisfies the manner of presentation character-
istic of the illusion, but that thing or individual feature is not itself perceived. (See
the case, below, of memory “filling in”.) The upshot is that either way, standard
illusions often do not have non-illusory counterparts, at least given our perceptual
systems as they actually are; but they do have veridical counterparts. This should
serve to drive home the distinction between veridicality and non-illusoriness.
To set out the point in some detail, we first need to distinguish genuine per-
ceptual illusions from cases where the subject is merely misled by what he or she is
experiencing. Here is a somewhat dated but potentially misleading presentation,
which is not a genuine perceptual illusion.
Of course, that is not a picture of Clinton and Gore, but a picture of “two
Clintons”, one with a Gore-ish haircut. If you initially saw it as a picture of Clinton
and Gore, your expectations caused you to misidentify the second figure. However,
this misidentification is not a genuinely perceptual illusion. It is an immediate
cognitive mistake, based on taking what is genuinely seen as a depiction of Clinton
and Gore. You took the figure depicted in the back to be Gore, because of mis-
leading clues and some things you genuinely saw, namely his Gore-like location at
the rear of the then President, and his Gore-ish haircut. Contrast this with genuine
perceptual illusions. Consider first:
This is not a case where you simply come to wrongly believe that the ball in
front is smaller than the ball in back. It is rather that the perceptual experience
itself presents relative sizes of the balls under a visual manner of presentation that
presents the ball in back as bigger. Indeed, if you are very familiar with the tricks
of perspective drawing you might have no tendency to believe the ball in the back
is bigger; you might make no mis-identification of the ball’s relative size, but you
still see it as bigger, and you also see the visible basis for this. Similarly with the
apparent curvature of the lines in the Hering Illusion; it is not just something we
are readily led to believe. There is a visible basis for that belief.
So also with the sine wave illusion; where in fact the lines are all of the same
length.
Now, consider the event of you attending to that array, an event which
involves the relative lengths of two lines being presented in such a way as to look
unequal in length. One question is: could an experience of that same type have
been veridical? If we individuate types of experience in terms of contents EX-ed
then it seems that an experience of that same type could have been caused, in a
non-standard way which by-passed some of the normal mechanisms of visual per-
ception, by an array which is just the way the array above appears to be. But even
so any such veridical version of the experience type in question would be illusory
in the sense that the subject would not be taking in the relevant visible features
of the array; she would not be aware of them as a result of the causal mechanisms
which typically operate in ordinary perception. For example, the visual system’s
The Problem with the Content View 117
FIGURE 5.7
Imagine that you call out the details of the display to your artist friend, who
follows your excellent directions exactly. You then look at the array he has pro-
duced and say “No. No. That is not it; the right vertical line is too long. Let us try
again.” No matter how many times you try, the right vertical line in the artist’s
model looks too long to match your original experience.
Your artist friend will be frustrated in just the same way in the case of the
Larger Ball Illusion, the Hering Illusion, the Smaller Center Illusion and the
Müller-Lyer Illusion. The perceptual mechanisms which generate these illusions
are such that their operations are constitutive of the particular acts of seeing the
scenes in question. That is why there are no non-illusory versions of such illusions,
at least absent serious rewiring of the human visual system.
Against Strong Content
118
One response on behalf of the content theorist is that resorting to genuinely singular
propositional contents can capture the still somewhat metaphorical idea of percep-
tual experience presenting, or putting us in touch with, the worldly items that entitle
us to believe the propositions we do about the scenarios before the senses.
So consider a phenomenally seamless transition from your hallucinating three
lights on in a ceiling, say as a result of the direct stimulation of your visual cortex,
to your then seeing each of the three lights on in the ceiling, say as a result of the slow
diminution of the direct stimulation to appropriate parts of your visual cortex,
while at the same time the lights in the ceiling are being turned up. At the begin-
ning of the seamless transition you were not “perceptually in touch with” the lights
on in the ceiling, at the end of it you were. One attempt to capture this difference
and thereby discharge the metaphor of presentation or perceptual contact is to
say that only at the end did you genuinely come to see those three lights on in the
ceiling. But this is not a report of a propositional attitude. How then does the con-
tent theorist model this transition in terms of a difference in the contents EX-ed at
the beginning and the end of the seamless transition?
One kind of content theorist, who supposes that the content of perceptual ex-
perience is always general (but for, perhaps, indexical reference to the subject and
the time of the experience), will not even attempt this, but instead will emphasize a
certain regular causal connection between the three lights on in the ceiling and the
EX-ing of the content that there are now before me three lights on in the ceiling.
If things are left there then the theorist is making nothing of the idea that our per-
ceptual experience itself puts us in touch with worldly items in a way that confers
a distinctive entitlement on what we typically go on to immediately believe on the
basis of that experience. For then, the seamless transition is not a transition in the
content EX-ed, and hence it is not a transition in perceptual experience itself, as
opposed to its external causes. It will then remain unclear why we are better placed
epistemically than a very reliable veridical hallucinator, who is never in experien-
tial contact with the relevant items in the scene before him.
Fortunately the content view is not, as such, committed to this kind of defla-
tion of the significance of non-hallucinatory experience. The content view can take
a different form. It can allow that the content EX-ed at the beginning of the seamless
transition is merely a general existential content to the effect that there are three lights
presently illuminated in a ceiling before one, while the content EX-ed at the end of the
transition is the singular content that those lights, that one, that one, and that one are
presently illuminated in the ceiling before one. In modelling that second content, the
lights themselves and the ceiling itself will be constituents of the proposition EX-ed.1
1
Perhaps they figure in the proposition under modes of presentation, in which case the singular
proposition will be in one way Fregean—in that it involves “de re senses”—rather than purely Russellian;
but this distinction does not matter for present purposes.
Against Strong Content
120
The resultant state of EX-ing may thus deserve the name of a distinctive form of con-
tact with the worldly items in question; for they enter into the individuation condi-
tions of the proposition that is EX-ed. The condition on the scene before the eyes
imposed by the experience so modeled is that those very lights be illuminated in that
very ceiling. Now the seamless transition from hallucination to genuinely successful
perception is modeled by a transition in the contents EX-ed. It can thus be claimed
to be a transition in what is experienced, and not just in how a given experience was
caused. What is experienced in successful perception thus goes beyond what is ex-
perienced in the corresponding hallucination precisely in respect of including the
worldly items actually seen or sensed.
The thing to notice is that even if this is an advance on the part of the content
theorist, one which may deserve the name of modelling how perceptual experi-
ence discloses, presents or “puts us in touch” with worldly items (the three lights,
the ceiling), the model does not explain how this kind of contact could distinc-
tively entitle one to go on to judge that there are three lights illuminated in the
ceiling. Of course, the belief this judgment lays down may be reliably formed, and
there may be nothing else one believes, or should believe, that counts against it.
But we should be holding out for a view on which the perceptual experience itself
plays a distinctive epistemic role in relation to immediate perceptual judgment.
Otherwise, sensory experience will end up being modeled as just a sensuous “light
show” merely accompanying the real epistemic transaction, as that transaction
is understood by the reliablist or, alternatively, by the theorist of prima facie jus-
tification. (For more on what is wrong about those pictures of the epistemic role
of experience see “On a Distinctive Epistemic Virtue” and the discussion of the
Wallpaper View in Johnston (2006) “Better than Mere Knowledge”.)
At the end of the seamless transition in question, the crucial thing is that you
are perceptually aware of the individual arrangement of those three lights in that
ceiling; your perceptual experience presents or discloses that individual arrange-
ment or “trope”-like entity to you. If you go on to judge that those lights are so
arranged in the ceiling then what you judged has what you experienced as its truth-
maker. The predicative structure of your judgment matches the complex exemplifi-
cation of features you experienced. Your experiencing the complex exemplifications
was not your enjoying some propositional attitude; to represent it as such simply
leeches out the distinctive epistemic contribution of experience.
There is a general lesson emerging here, one which will reappear in what fol-
lows. The content theorist who favors Russellian content has still failed to model
perceptual experience as putting us in touch with the worldly items in a way that
entitles us to immediately believe the propositions we do about the scenarios before
the senses. To the extent that the content theorist incorporates the worldly items
sensed into the propositional content EX-ed, he or she may capture the idea that
those items genuinely figure in the experience, but the epistemic significance of
the items figuring in the experience is then lost. This is because experience of the
The Problem with the Content View 121
items in question is not correctly modeled by those items simply being subjects of
predication.
In successful perceptual experience we are aware of exemplifications of per-
ceptual features, and not simply of predicative relations among items and features.
The exemplifications can play an epistemicially significant role that no mere pred-
ication of perceptual features of perceived items can play. For the exemplifications
are the truthmakers of the corresponding predications, and hence of what we go
on to immediately judge. Once we recognize this, there will be no need to theo-
retically posit a propositional attitude like EX-ing that p. Perceptual experience is
directed at exemplifications (under certain modes of presentation), or in the case
of illusion and hallucination, merely ostensible exemplifications (under certain
modes of presentation).2
In contrast, perceptual judgment and the beliefs laid down by perceptual
judgment do involve predications of perceptible features of perceptible objects.
Perceptual experience is epistemically relevant not because it is a “prehearsal” in
the sensory system of such predications, but because it discloses truthmakers for
what we immediately judge on the basis of perception.3
(It should be noted that this claim does not involve a commitment to the so
called truthmaker axiom to the effect that there are no truths without truthmak-
ers, any more than the mention of tropes commits one to trope nominalism as a
solution to the problem of universals.4)
Another example may serve to further illustrate the idea of successful perception
as the disclosure of truthmakers. Consider a seamless transition from a case where
memory fills in a feature that is not in fact sensed, to a case where the feature is
genuinely sensed. Perhaps memory fills in low-light vision, so that in the gloaming
or twilight we have a more vivid appearance of the colors of just those objects that
are familiar to us. So imagine you are looking into the backs of cars when dusk has
settled. (Don’t ask!) With unfamiliar cars you see the leathers in shades of grey.
But with your own car, memory fills vision in, and you see your leather not in the
grayscale but in the red scale. You don’t however see the redness of the leather.
Memory’s function here is, we may suppose, not part of the normal functioning
of the visual system itself, and so it is not a way of seeing the individual colors
of things. So when things work like this we have a case of veridical illusion; you
2
In “The Obscure Object of Hallucination” (2004) I called such ostensible exemplifications
“sensible profiles”. There I failed to emphasize that even such ostensible exemplifications are given
under modes of presentation.
3
For a similar view, derived in part from Edmund Husserl, see Kevin Mulligan 1995.
4
For a general theory of tropes and of trope nominalism as a solution to the problem of universals,
see Johnston (1983) Particulars and Persistence (Princeton University, Ph.D.)
Against Strong Content
122
don’t see the redness of the leather, you are only aware of an expanse of redness,
which is not the individual redness of the leather. But then there is a transition,
not detectible by you; either the discriminative capacity of your rods is suddenly
augmented, or a discrete unnoticed light goes on in the car, so that you then do see
the redness of the leather.
The thought is that this transition, just like the seamless transition from hal-
lucinating lights to seeing lights, is potentially epistemically significant in that, at
the end, you have one sort of exploitable entitlement to judge that the leather is red,
one that you did not have at the beginning. At the beginning, you may well have
been prima facie justified in believing that the leather is red, in that it seemed to
you to be that way, and no other belief you had or should have had was at odds
with this. Moreover, memory’s filling in perception during the gloaming may even
be a reliable process, so that even at the beginning, your immediate belief that the
leather is red was reliably formed. Still, at the end, but not at the beginning, you
were visually aware of the redness of the leather, and this itself provides an exploit-
able entitlement to go on to judge that the leather is red. (As to just how we exploit
the entitlement, see “On a Neglected Epistemic Virtue”.)
The source of that entitlement is entirely occluded if we model your percep-
tual experience at the end—that is, your awareness of the redness of the leather—
as a propositional attitude with the content that the leather is red. For once again
this would trade in your awareness of an exemplification for the having of a prop-
ositional attitude directed upon the corresponding predication. Thereby, we lose
precisely the distinctive sort of thing which successful perception discloses or
presents.
Notice that that the source of the entitlement is also occluded even if we go for
more arcane propositions as the contents that are EX-ed. There is no good reason
why tropes like the redness of the leather cannot be themselves subjects of predi-
cation. We can predicate existence of them, we can predicate location of them, we
can make comparisons concerning them, and so on and so forth. But the proposi-
tion that the redness of the leather exists is not what makes that proposition true,
it is the redness of the leather that does this, and that is why enjoying a perceptual
experience that discloses the redness of the leather is epistemically relevant to the
belief that the redness of the leather exists.
It is at this juncture that I seem to hear John Searle in the wings, saying “I told you
so!” (Sadly, it is, as far as I can tell, only a hallucination on my part.) For Searle’s
original version of the content view was precisely designed to close the gap be-
tween an experience’s being veridical in the sense of saying something true about
the subject’s environment and its being non-hallucinatory and even non-illusory.
Once we left the ambition to close that gap behind, we opened up space for the
The Problem with the Content View 123
kind of cases I have been discussing, namely veridical hallucination and veridical
illusion, where the non-Searlean content theorist seems unable to properly locate
the relevant defects at the level of the content EX-ed, and so cannot explain why
the transitions in question to the “fully successful” perceptual experiences might
be improvements from the epistemic point of view.
(Remember, we are not trying to answer skepticism here, so we are not
making the mistake, often execrated by Searle, of trying to condition our account
of perception by the need to answer skepticism. We are instead trying to explain
how, skepticism aside, our being fully successful experiencers, in particular our
not being merely reliable victims of hallucination or illusion, could confer upon
us some distinctive epistemic advantage. It is a bit of a scandal if we have nothing
interesting to say about that.)
So we should look again at Searle’s own version of the content view, as it
applies to the case of perceptual illusion in which memory fills in colors. In that
case, there is a transition from seeing the leather but not the redness of the leather,
to seeing the redness of the leather, thanks to augmented rods or better lighting.
Now Searle has a general argument that a perceptual experience involves more
than characterizing its target in terms of how it is intrinsically and in terms of its
relations to the objects around it. Perceptual experiences also characterize their
targets as causing the very experiences in question. Searle motivates this by way of
a contrast between the phenomenology of genuine perception and what he plau-
sibly supposes might be the phenomenology of a perfect eidetic imager. Searle
holds that such a person would not have the sense of being passively under the
influence of the target, as we are in standard cases of perception. The passivity of
perception is part of the experience of perception itself, and Searle proposes to
capture this fact by supposing that the content of perception mirrors this. If one
is seeing a yellow station wagon over there then the content of the perception is
not just that there is a yellow station wagon over there but also that it is causing
the very experience in question. In this way, Searle hopes to close the gap between
perceptually entertaining a veridical content and having a fully successful percep-
tual experience.
So in the case where memory is filling in, the content to the effect that the
redness of the leather is (in part) causing the very experience in question might be
taken to be false at the beginning, while at the end of the transition the content to
the effect that the redness of the leather is (in part) causing the very experience then
in question is true. Is that why one is better placed epistemically at the end of the
transition?
Several remarks are in order here. In the case at hand, memory “fills in red-
ness” on the basis of remembering the redness of the leather. On any reasonable
account of memory, the state of remembering the redness of the leather is (in part)
caused by the redness of the leather, and so given that, in this kind of case, a cause
of a cause of a thing is also going to be a cause of the thing it follows that, con-
trary to the suggestion above, the content that the redness of the leather is causing
Against Strong Content
124
this very experience will be true during the experience that involves memory
filling in. This is in effect the observation that, as stated, Searle’s account of the
self-referential causal content of experience does not in fact close the gap between
veridical experience and fully successful experience. Precisely because of wayward
causal chain cases like memory’s filling in, we have to add that the causation be
of the right sort, namely of the sort that is typical of fully successful perception.
But this addition to the content of each perception is implausible; the need for
“causation of the right sort” was discovered by sophisticated thinkers late in the
philosophical day—I suppose in the early 1960s—in response to proposed causal
analyses of action and perception, and not by attending to how the world seems to
be in perceptual experience. In any case, the required patch-up for Searlean content
goes far beyond what Searle motivates by way of noting the sense of passivity—the
sense of being under the causal influence of the target—in sensory experience.5
However, let us put that somewhat familiar objection aside for now. Suppose
we allow that seeing the over-there-ness of a yellow station wagon and collaterally
sensing its causing the very experience in question reduces to having a perceptual
experience to the effect that there is a yellow station wagon over there and that the
yellow station wagon is causing the very experience in the right way. It still seems
that this version of the content view makes a form of the mistake that I earlier
suggested was characteristic of the content view. It begins with objectual attitudes
directed at “tropes” or individual characteristics and it translates those into prop-
ositional attitudes directed at the corresponding predications, i.e. the predication
of a certain location of the car and the predication of causing experience of the car.
These are not only very different things that are being assimilated by the trans-
lation, but the translation again occludes what is epistemically interesting about
perceptual experience, namely that it involves the disclosure of truthmakers for
what we are inclined to go on to judge. It is one’s seeing the relative location of the
yellow station wagon that provides an exploitable entitlement to judge that the
station wagon is over there. It is one’s experiencing the yellow station wagon as
producing our experience that provides an exploitable entitlement to judge that
the yellow station wagon is producing one’s experience.
So the most revealing objection to Searle’s distinctive version of the content
view is just that is it a version of the content view, and thereby occludes the epi-
stemic significance of perceptual experience; that is, its capacity to disclose truth-
makers for what we typically go on to immediately judge.
So far, in arguing that the content view occludes the significance of the percep-
tual presentation of truthmakers, we have focused on cases where the truthmakers
5
A similar point is made by Millar (1985).
The Problem with the Content View 125
conclusive entitlement to believe or judge that that your visual field is filled
with an expanse of Brain Grey*.
The same considerations go through even for those of us who believe that
there are no visual fields (Johnston 2011b). In that case, what one is aware of in
being aware of Brain Grey* is an expanse of Brain Grey* appearing before one.
The content theorist will model this as an EX-ing of a proposition to the effect that
an expanse of Brain Grey* is appearing before one. But once again this obscures
the contribution of perceptual experience. The perceptual experience in question
was directed at a truthmaker for that proposition. As such it puts the subject of the
experience in a position to exploit a conclusive entitlement to believe or judge that
an expanse of Brain Grey* is appearing before him or her.
(These same points could be made with after-images, and with expanses of
“film color”, a neglected but extremely interesting phenomenon, of considerable
import in the philosophy of perception.)
So far we have seen that the content view (i) mischaracterizes the success conditions
of experience, and (ii) occludes the distinctive epistemic significance of experience
by modelling it terms of a propositional rather than an objectual attitude.
If we simplify and treat hallucination as a certain kind of extreme illusion,
namely the kind in which there is an illusion of a particular individual, event or
quantity of stuff being presented to us, then we can say that experience is bipolar
in that it does come in two forms or types, namely illusory and non-illusory
types. But this “bipolarity” of experience is not captured by the truth or falsity
of propositions EX-ed. That dimension of contrast merely serves to highlight
the veridicality or non-veridicality of experience, and as we have seen there can
be veridical illusions, i.e. illusions in which we are not presented with the truth-
makers of what experience naturally prompts us to judge, even though those
truthmakers are there to be experienced in the scenarios before the senses, but
are nonetheless missed because of perception failing us, as in the case of mem-
ory filling in. The lesson is that if we are to characterize the illusory/non-illusory
distinction, or more generally the hallucinatory cum illusory/non-illusory cum
non-hallucinatory distinction, we need to recognize that fully successful percep-
tual experience involves objectual attitudes directed at truthmakers.
This is the same conclusion that is made plausible by the consideration that
the content view systematically mischaracterizes the kind of entitlement that per-
ceptual experience puts us in a position to exploit. That kind of entitlement is
typically conclusive, at least for the simple immediate perceptual judgments that
we habitually make.
The two routes to the same conclusion are obviously related. For if you mis-
characterize the illusory/non-illusory distinction as the non-veridical/veridical
The Problem with the Content View 127
distinction you will as a result fail to appreciate just what the epistemic import of
perceptual experience is.
Given this description of the terrain, those trying to play out the argumentative
chess game on behalf of the content theorist might now make the following move:
Right; we content theorists will now accept there are two ways in which per-
ceptual experience can go wrong. It can be non-veridical and it can be illusory.
The existence of veridical illusions and veridical hallucinations shows that, but
the proper response to this fact can be accommodated within a modified con-
tent view.
First Attempt (Conjunctivism): EX-ing that p, we forgot to tell you, comes
in two forms, namely luding that p and the more demanding form of percep-
tually taking in the fact that p. A case of luding that p is a mere case of EX-ing
that p. It is veridical just when p obtains, but even when luding is veridical it
falls short of what is involved in perceptually taking in the fact that p. Your
perceptually taking in the fact that p involves your EX-ing that p being caused
in the right way by a truthmaker of p there in your perceived environment.
Second attempt (Disjunctivism): “EX-ing that p”, we forgot to tell you,
is an overarching term for two quite different kinds of states, namely lud-
ing that p and perceptually taking in the fact that p. Beware: states of the
second are not special variants on the states that make up ludings; they are
not appropriately caused ludings, they are sui generis states. A case of luding
that p is veridical just when p obtains, but even when it is veridical, luding
falls short of what is involved in perceptually taking in the fact that p. Your
perceptually taking in the fact that p involves the underlying functioning of
your visual system being caused in the right way by a truthmaker of p there
in your perceived environment, but it is more than that. It involves the pres-
entation of the fact that p.
Now we can explain the distinctive epistemic significance of “fully suc-
cessful” veridical experience, i.e. perceptually taking in the fact that p. If
you perceptually take in the fact that p, you are aware of a truthmaker for
what you go on to immediately judge on the basis of perception. You have
an exploitable entitlement to judge that p, one that differs from the reli-
ability of the process leading to that judgment and from any prima facie
justification that attends that judgment.
A number of remarks need to be made here. First, the conjunctivist version of the
revised content view does not in fact model perceptual experience as putting us in
touch with a truthmaker for p; at best it models experience as relating us to true
propositions, and not to their sensed truthmakers. On the conjunctivist version,
those truthmakers come in simply as the causes that distinguish the good case of
fully successful perception from mere true luding that p.
It is the disjunctivist version, which is reminiscent of John McDowell’s (1996,
1998) account of perception, which delivers perceptual awareness of truthmakers
Against Strong Content
128
the single malt is peaty are not at all remarkable to her, so neither is more remark-
able than the other. Still, she could find the astringency of the calvados and the
peatiness of the single malt both remarkable, and the astringency of the calvados
even more remarkable that the peatiness of the single malt. She could make such
comparisons on the basis of what she experienced (and her background knowl-
edge.) So the astringency of the calvados and the peatiness of the single malt
should be counted among the things she experienced.
The calvados’ particular astringency which Mary tasted could not continue to
exist without maintaining its utterly determinate character, and the single malt’s
particular peatiness which Mary tasted could not continue to exist without main-
taining its utterly determinate character. Yet, the fact that that the calvados is astrin-
gent could continue to obtain while the calvados’ particular determinate degrees
of astringency varied, and the fact that the single malt is peaty could continue to
obtain while the single malt’s particular determinate degrees of peatiness varied.
Facts register the obtaining of certain conditions, without thereby registering the
more determinate details of how they obtain. But experience takes in the tropes or
individual features of things, and those features are individuated by the determinate
way they are. It is by taking in tropes or individual characteristics that experience
provides for comparisons of the sort Mary is reported as making when we say
Mary found the astringency of the calvados more remarkable than the peati-
ness of the single malt.
The friend of facts may see a way forward here, by building in a specific determi-
nateness into the facts perceptually taken in by Mary. If these factive attitude reports
were true
Mary perceptually took in the fact that the calvados was astringent-to-degree n.
Mary perceptually took in the fact that the calvados was astringent-to-
degree m.
Mary would have been put in a position by her experience, as it is here modeled,
to find the degree of astringency of the calvados more remarkable than the degree
of peatiness of the single malt. At least she would be, if as we might suppose, she
is aware of the common degrees of astringency and peatiness of the respective
spirits. (The suggestion would have to be elaborated for any dimension of com-
parison along which Mary might find one of the respective tastes more remark-
able than another. For example, Mary may in fact be finding the astringency more
remarkable because of its purity, not its intensity.)
But this suggestion runs afoul of the fact that our perception of magnitudes,
as opposed to our subsequent theorizing about them, is unit free. “Degree n”
and “degree m” are only given sense in a theory of measurement or comparison,
and that is a late achievement of cognition, not part of the everyday deliverances
of perception. Moreover, we can experience determinate individual features
which admit of comparison in respect of their degrees along some dimension,
Against Strong Content
132
without thereby experiencing just what their degrees along that dimension are.
But the factive paraphrase, employed to solve the problem of Mary’s compar-
ison, appears to place the degrees themselves in the facts experientially taken in.
Thus, reverting to facts taken in rather than individual characteristics pre-
sented invariably misdescribes what we are given in experience. And this is just
the beginning of a cascade of points against the idea that the right way to parse
our experiences is as the “perceptual taking in” of facts. Indeed we need to resort
to that unfamiliar idiom precisely because facts are not seen, or heard or smelt or
touched or tasted. We might eventually arrive at a relation to a fact on the basis of
what we see, hear, smell, touch or taste—we might see that the flag is blue—but
that is a destination from, and not a starting point in, what we are aware of in
experience.
For these reasons, among others, the objectual reports involving Paul, Jane,
Sam, Mary and Fred cannot be adequately paraphrased either in terms of proposi-
tional or factive attitudes. The ASES that the objectual reports describe need to be
recognized in any account of experience as we ordinarily conceive of it. The ASEs
involve the presentation of objects, quantities of stuff, events and tropes under
certain manners of presentation. The items presented are truthmakers for the
simple existential and predicative judgments we immediately make on the basis
of perception.
Recognizing the ASEs also enables a response to a pressing worry faced by recent
philosophical discussion of experiences, the worry that much talk of experiences
is itself too unspecific to be of real philosophical use (Travis 2004). There is the
quasi-mass noun use of “experience”, as in “experience is mostly veridical”. There is
the dummy sortal use, as in Alex Byrne’s example—“My three most embarrassing
experiences in graduate school”—where we can paraphrase away apparent ref-
erence to three token experiences in favor of three occasions of embarrassment.
Travis and Byrne urge that philosophers need an account of experience that goes
beyond the thin commitments of such usages, at least if the notion of an experi-
ence is to do any real philosophical work.
Hence the relevance of the ASEs; they are paradigmatically experiential, they
are countable and datable episodes, and they are taken for granted by, and studied
in, perceptual psychology and psychophysics (Johnston 2011a). Moreover, we do
in fact report such episodes in describing our sensory interactions with our envi-
ronment. Furthermore, we can use attentive sensory episodes to understand the
notion of a stretch or course of experience, i.e. the total experiential state of a sub-
ject as it evolves over a given period of time.
I open my eyes, and something happens; I enjoy a course of visual experience,
which involves and supports a range of token ASEs. Nothing could be that very
The Problem with the Content View 133
course of experience unless it involved and supported the same token ASEs. As
the lighting or perspective changes, and as my attention-driven gestalt groupings
change, the course of experience changes its character. A course of experience is
thus partly constituted by a series of ASEs, with one or another evolving out of pre-
vious ASEs in accord with changes in the subject and the environment.
For the reasons already canvassed we can’t fully explain the nature of courses
of experience in terms of successions of propositions EX-ed or facts taken in; for
both sorts of successions omit, each in its own way, the presentational character
of the ASEs that are at least in part necessary to a course of experience being what
it is.
True, not everything given in a course of experience is an object of attention,
and hence a target of an ASE, but if something is experienced it is a potential ob-
ject of an ASE (residual elements aside). So let us individuate token courses of
experience in terms of the ASEs they ground, or actually and potentially support.
Then we can individuate types of experiences by abstracting away from the sub-
ject, particular constitution and time of occurrence of token courses of experi-
ences. We can say:
A type of course of experience E = E* if and only if (subject, token constitu-
tion and time aside), E and E* have the same actual ASEs (same targets, same
manners of presentation) in the same order and the same ground for potential
ASEs, and the same residual elements, all in the same sequence and pattern.
Among the residual elements which enter into a course of experience may be
forms of unconscious perception, which shape the character of what is perceived
both at the level of objects presented and their manner of presentation. There may
also be ASEs that are themselves unconscious in the sense of not being available
in introspection, and these also may shape the conscious character of a course of
experience (Block 2011). The important point is that such residual elements need
not be modeled in terms of propositions EX-ed or facts perceptually taken in.
In any case, all we need to rely upon in what follows is a necessary condition
on being a particular course of experience, namely:
E = E* only if (subject, token constitution and time aside) E and E* have the
same actual ASEs.
Since there is no modeling of courses of experience without modeling the ASEs,
it follows that if ASEs cannot be modeled adequately by propositional attitudes or
facts taken in, then neither can courses of experience.
ASEs involve attention, attentive search and the like. Although in some cases ASEs
may be forced on us—the thud may be so loud that it captures our attention—often
Against Strong Content
134
FIGURE 5.8
they are things we do because of explicit beliefs and desires that we have. We can
be ordered or urged to attend. “Try to visually discriminate the depiction of the
young woman in the array!” Just because ASEs are early on in perceptual proc-
essing doesn’t mean they are not directed by what is already there “later on”.
Thirsty people may be better at visually detecting transparency. And there is some
evidence that emotional attachment effects size perception. Moreover, we are often
enjoying sensory episodes as part of an explicit investigation of our environment,
as in a search for clues.
Even so, undergoing an ASE in which a certain object is presented as being a
certain way does not intrinsically involve one in taking the thing to be that way.
Someone very familiar with the following illusion and someone entirely unfa-
miliar with it could each have an ASE of the very same type, one in which the
pencil is presented as bent. The first person will not go on to believe that the pencil
is bent, while the second is likely to do so.
In presenting the pencil as bent, the relevant ASE is not silent as to how the
world might be, but in enjoying that ASE the subject him-or-herself is not com-
mitted to the world being a certain way. Experience, understood in terms of ASEs,
is not neutral as to the way the world is, but undergoing an ASE does not commit
The Problem with the Content View 135
the subject to the world being a certain way, and certainly not in the fashion in
which the subject’s judgments or beliefs do.
In this way we can make sense of the idiom of “the testimony of the senses”.
Experience involves a commitment to how the world is, but it is, as it were, experi-
ence’s own commitment, which a subject can take on as his or her own by trusting
experience, i.e. coming to adopt the beliefs that experience supports, the beliefs
whose truthmakers are presented in experience, under recognizable modes of
presentation.
For many friends of the content view, the theoretically posited EX-ings
that p were intended as experiential attitudes that were prior to perceptual
judgment and belief, not only in time, but also in the sense that although
EX-ings had their own conditions of veridicality and hence themselves in-
volved “a commitment” to the world being a certain way, the EX-ings could be
enjoyed by subjects without those subjects being committed to the world being
a certain way.
That role—the role of characterizing the kind of commitment character-
istic of experience, in virtue of which it can come in veridical and non-veridical
forms—can also be played by the ASEs. But as our sequence of standard illusions
illustrated, we do better in this regard by relying on the ASE’s; we can explain how
the commitment characteristic of experience allows for a further kind of failure
exhibited by illusions which are nonetheless veridical.
So EX-ings are not only inadequate models for the ASEs; they are redundant
and not fully adequate as models of the kind of commitment incurred by experi-
ence. Moreover, they are theoretical posits, rather than episodes we actually report.
Finally and most importantly, unlike the ASEs, the EX-ings do not model experi-
ence in such a way as allow for the distinctive epistemic virtue that attends our im-
mediate perceptual judgments; those judgments are formed in the sensed presence
of their truthmakers, and this provides us with a distinctive kind of entitlement to
those perceptual judgments, an entitlement which we can produced if challenged.
What gives one the right to believe that the leather is red? Nothing could serve
as a better response than saying “I see the redness of the leather.” If that latter claim
is true, the subject possesses and is thereby exploiting a conclusive entitlement for
his belief. His belief couldn’t but be true given what he experienced. Saying “I am
EX-ing that the leather was red” describes no such entitlement. The proposition
that the leather is red leaves open whether it is true or not, and so leaves open
whether the belief that the leather is red is true or not. Successful perception actu-
ally closes that option off.
The standard version of the content view, which has EX-ing that p come before
judging or believing that p, faces the old question of how something that is not a
Against Strong Content
136
belief can nonetheless bear on what one should judge or believe. Awareness of a
truthmaker for p can bear on the judgment that p, in that it is a presentation of an
exploitable entitlement to judge that p. But how does some propositional attitude
that is not itself the belief that p, indeed some attitude that does not itself involve
the subject in taking p to be true, bear on whether the subject should judge or be-
lieve that p?
It is noteworthy that Byrne is here led to cut the Gordian knot, and propose
that EX-ing is a kind of believing, the kind of believing where the particular sorts
of contents believed are those characteristic of perceptual experience, and not
those that enter into the corresponding verbal judgments. That is, the believings
that are EX-ings have contents restricted to the predication of visual gestalts made
up of “mid-level perceptual features” such as shape, size, motion, color, shading,
texture, orientation, timbre, loudness, pitch and the like, the very kinds of features
perceptual illusions prompt us to misattribute to objects.
Of course, Byrne is well aware of the fact that in cases of known illusion such
as the illusion of the bent pencil in water, we are not inclined to believe what the il-
lusion seems to show, i.e. that the pencil is bent. Byrne returns to the view of David
Armstrong (1968), suggesting that in such cases the belief in question is prevented
from playing its usual causal role because of the other things we believe, namely
that the presentation of the bent stick is an illusory appearance of the pencil’s shape.
Whatever other difficulties attend this view6, the main point to make is that
this is just another way of giving up on the distinctive epistemic contribution of
fully successful perceptual experience. Our immediate perceptual beliefs are often
reliably caused. But beyond that there is little to be said on a view like Byrne’s about
the distinctive epistemic significance of perceptual experience. Someone who had
reliably caused non-experientially based beliefs about the scenarios before the
senses—say a perfect blind-sighter and deaf-hearer—could be as well-placed epi-
stemically as those of us who can see and hear.
A theory’s yielding that implication is the definitive sign that the theory is
ignoring the distinctive epistemic significance of experience. It now appears that
all varieties of the content view do this in one or another way.
The “content of experience” frame that has shaped the recent discussion of
perception over the last thirty years is far from innocent; instead it is demonstrably
misleading, and it is disabling when it comes to providing an account of the dis-
tinctive epistemic significance of perception.
References
6
Does it conflate an experience’s commitment or “testimony” with a commitment of the
experiencer?
The Problem with the Content View 137
Chisholm, R. (1957). Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press.
Crane, T. (2006). Is there a perceptual relation? In T. S. Gendler & J. Hawthorne (Eds.),
Perceptual Experience (pp. 126–146). Oxford: Oxford University Press.
Crane, T. (2011). Is perception a propositional attitude? In K. Hawley & F. Macpherson
(Eds.), The Admissible Contents of Experience (pp. 83–100). Oxford: Blackwell.
French, C. (2013). Perceptual experience and seeing that p. Synthese, 190, 1735–1751.
Harman, G. (1973). Thought. Princeton: Princeton University Press.
Hyslop, A. (1983). On ‘seeing-as’. Philosophy and Phenomenological Research, 43(June),
533–540.
Jackson, F. (1977). Perception. Oxford: Oxford University Press.
Johnston, M. (1983). Particulars and Persistence. PhD thesis, Princeton University.
Johnston, M. (1992). How to speak of the colors. Philosophical Studies, 68(3), 221–263.
Johnston, M. (1997). Postscript: visual experience. In Alex Byrne & David Hilbert (Eds.),
Readings on Color I: The Philosophy of Color (pp. 172–176). Cambridge, MA: MIT Press.
Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 120(1–3),
113–183.
Johnston, M. (2006). Better than mere knowledge? The function of sensory awareness.
In T. S. Gendler & John Hawthorne (Eds.), Perceptual Experience (pp. 260–290).
Oxford: Oxford University Press.
Johnston, M. (2007). Objective mind and the objectivity of our minds. Philosophy and
Phenomenological Research, 75(2), 233–268.
Johnston, M. (2011a). On a neglected epistemic virtue. Philosophical Issues, 21(1), 165–218.
Johnston, M. (2011b). There are no visual fields (and no minds either). Analytic Philosophy,
52(4), 231–242.
Johnston, M. (n.d.). Ontology without Properties. Manuscript.
McDowell, J. (1996). Mind and World. Cambridge, MA: Harvard University Press.
McDowell, J. (1998). The Woodbridge Lectures: Having the world in view: Sellars, Kant, and
intentionality. Journal of Philosophy, 95, 431–491.
Millar, A. (1985). Veridicality: More on Searle. Analysis, 45(March), 120–124.
Moltmann, F. (2013). Abstract Objects and the Semantics of Natural Language. Oxford: Oxford
University Press.
Mulligan, K. (1995). Perception. In B. Smith & D. Smith (Eds.), Husserl, Cambridge
Companions to Philosophy (pp. 168–238). Cambridge: Cambridge University Press.
Searle, J. (1983). Intentionality. Oxford: Oxford University Press.
Siegel, S. (2006). Which properties are represented in perception? In T. Gendler & J.
Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford University
Press.
Travis, C. (2004). The silence of the senses. Mind, 113(449), 57–94.
6
Have we made the progress McDowell speaks of? Science has, no doubt. In philos-
ophy, though, that mediæval idea may masquerade as science itself. Masquerade
only: nothing, so not science, suggests messages in nature of the sort found in
books. So I will argue. Nature is full of messages for us. That red sky at night tells
a sailor something. But it is superstition to approach such messages as one would
a text or utterance or speech act—though superstition which still tempts some.
What distinguishes the messages in texts or speech acts? First, they are issued,
produced, conveyed, by some author. (They are also borne by, or contained in, the
text or act itself.) Representing can just be holding a stance or posture towards
things, a condition one is in. Representing something to be so (henceforth
representing-to-be), e.g., can just be taking it to be so. I will call such represent-
ing autorepresenting. Such will be a side issue here. By contrast, the authoring of a
A slightly different version of this essay also appears in my, Perception: Essays After Frege
(Oxford: Oxford University Press, 2013). I thank OUP for permission to print it here. I am grateful to
Mike Martin, Mark Kalderon, Guy Longworth, and Craig French for helping me see where some of
138 the lines here lead.
The Preserve of Thinkers 139
6.1. Thinkers
The notion of a thinker at work here is Descartes’. Aiming to distinguish res cogi-
tans from dumb brutes and refined machines, he offers two marks,
of which the first is that they [machines, brutes] could never use words or
other signs, composing them as we do to express their thoughts to others.
For one could indeed conceive of a machine being so arranged that it offered
words, and even that it offered certain ones about material actions causing
certain changes in its organs . . . but not of it arranging them diversely so as to
respond to the sense of all that was said in its presence in the way that even
the most mentally deficient men can. . . . And the second is that, while they did
several things as well as, or perhaps better than, any of us, they would infal-
libly fall short in others, by which one would discover that they did not act
through knowledge, but solely by the disposition of their organs. (1637, p. 92)
“Reason,” Hilary Putnam wrote, “can transcend whatever it can survey” (Putnam
1988, p. 119). Such is Descartes’ idea. Take any implementable theory of how to
The Preserve of Thinkers 141
thought that, but mistakenly. Another possibility. If the wild boar were not in rut,
things would not be as Pia thinks. Do what you like, and but for that fact Pia’s
picture of the world would not jibe. But an overly genteel upbringing has left her
without the notion for a beast to be in rut. Perhaps she might still count as thinking
that the wild boar are in rut; but not thanks to her ability to identify that as a way
she thinks things. Perhaps (for all that matters here) a cat might stand similarly
towards a hole’s presence in a wall. Such, anyway, are reasons for separating two
notions of thinker as I have just done.
6.2. Generality
‘that’ attached, their so being as enjoying a certain status: as part of how things are.
Representing-to-be takes a further step: not merely presenting a given way as being
a way things are—what would be just more representing-as—but as assuming, or
incurring, liability to a particular sort of success or failure, getting it right or wrong.
No thought takes this extra step. It cannot aim at such success or failure (or any-
thing). The thought that pigs swim is not to blame if they do not.
Whence this generality? Following Frege, a thought is what brings truth
into question at all, done only by fixing (or being) a particular question of it; a
particular point on which thinkers might agree or not. One cannot simply aim
at truth tout court. It must be truth in re something. Which is to say: one cannot
aim at everything. So a question of truth cannot turn on everything. Whether that
Mondriaan is Dutch may turn on Mondriaan’s parentage, but not on whether
Pia was at Hédiard yesterday, or Sid is wearing sandals. It follows that a range of
cases—an indefinitely large one—are ones which would, or might, count as things
being such that that painting is Dutch—ones with Sid in socks and sandals, ones
with him pieds nus, and so on ad inf. A thought (and that way for things to be
which it is of), reaches in its own way to particular cases, thus reaching just what
it does. How it reaches is contained in it being the thought it is. Thus a thought’s
inherent generality and that of a way for things to be.
Frege puts two pieces in play. Thoughts, so ways for things to be, are one piece.
The other is what he calls ‘the particular case’—what a thought presents as falling
under some generality. What falls under a generality is intrinsically one-off: nothing
else could be things being as they now are. What makes the particular particular,
though, is rather its lack of reach. Nothing in its being the case it is identifies any
question of truth, or what matters to it. The sun is setting slowly over the Douro’s
mouth. For the sun to be setting slowly is a way for things to be. Things being as they
now are is a case of this. Study that case as closely as you like, and you will not learn
from it what matters, and how, to whether a particular case would be a case of this or
not. For this one must look at just what generality is to be instanced. Generalizing,
no proper part of a generality’s reach determines what further reach it might or
might not have. Generalities and particular cases are thus two fundamentally dif-
ferent sorts of things. I will speak of the first, ways for things to be, as conceptual, the
last, things being as they are, as nonconceptual. That core relation between these two
domains, being a case of, I will call instancing, its converse reaching to.
One can witness, e.g., watch, things being as they are. One does this, e.g., in
seeing the sun, setting over the Douro’s mouth. What is visible—the sun, e.g.,—has
location. What has location is what may interact causally with its surroundings.
Such is part of Frege’s point in insisting that thoughts cannot be objects of sen-
sory awareness. They are the wrong sorts of things for that. They are equally unfit
for causal interaction. It cannot be the causal profile of a way for things to be
which makes it occupy the third place in the relation allorepresenting for given
first and second terms. It has no such profile. Allorepresenting cannot be made of
effect-representing by any such route.
Against Strong Content
146
a theory which generated in advance all the reasons there might be (or might
have been) for this. Nor, correlatively, could one expect to say in advance what
it would mean (factively) for the Porsche to be yellow if its failure here did, and,
again, if it did not, cancel its claim to count as being yellow—as instancing the
generality being a yellow Porsche. So nor could there (plausibly) be a theory
which predicted in advance, when it would be true to what a Porsche’s being
yellow is, where there was such a failure, to rule in the one way or the other.
Which, if right, is to say: there can be no specifiable prosthetic for our sense
of when to say (when it would be true to say), when not, that a Porsche is yel-
low. Which is to say: the ability to see this draws essentially on those capacities
which mark a Cartesian thinker. Thus, too, for the ability to see when things
would be as Sid represented them in representing Pia’s Porsche as yellow, so
the ability to grasp what way his created way—things being as he represented
them—is.
An ability to see what would, or might, count as a case of something being
yellow is very different from a mere ability to detect what are in fact cases of what
does so count, as an ability to see what would count as something being a pig dif-
fers from an ability to tell a pig at sight. An ability to tell a pig at sight is that thanks
to the fact that pigs are recognizable by certain visual features—by how they look.
But we all recognize that to be a pig is not, certainly not just, to have those features.
Not all that grunts is, or need be, porcine; not all that is porcine need grunt. So
an ability to tell pigs at sight is that only in a hospitable environment. Flood the
environment with enough ringers, and it ceases to be an ability at all. An ability to
see what would count as something being a pig transcends such limits. It is, inter
alia, an ability to see when we have ringers to deal with. Such an ability is what is
drawn on in identifying what way Sid represented things being in representing
Pia’s Porsche as yellow. It is for such abilities, I have suggested, that there is no
prosthetic.
An author of allorepresenting is responsible for his creations. Blame for suc-
cess or failure—e.g., for representing things as they are not, or as they ought not
to have been—is to be lain at his door. But he can be blamed only for what is in his
control, for what he/it can be responsive to having done or not. He could, might,
have done otherwise; he is thus blameworthy for not having so done. The point
just made is, in brief, only a Cartesian thinker could be thus responsible for having
represented things in one way rather than another.
To some this will seem wrong. To their eyes, nature, or some of its creations,
though no thinker in any sense, can assume the sort of responsibility for some
of its (or their) productions that one does per se in allorepresenting; notably the
sort of responsibility one does in representing truly or falsely—in making oneself
liable for being right or wrong as to how things are. The rough idea is: those cre-
ations exist to fulfil a purpose; they assume the responsibility something would in
undertaking to fulfil that purpose. What follows, I hope, will demolish that idea.
Against Strong Content
148
6.3. Selecting
selection. A Cartesian thinker guides his projects by his perceptions of how the
world bears on what the thing to do would be—on which projects to execute, and
how. His perceptions reflect an unbounded sensitivity to ways the world does, and
might, bear on this. So it is, in particular, with his perceptions of what to allorepre-
sent, and how. We represent with an agenda. Such agenda may include contribut-
ing to, or furthering, further projects in particular ways—e.g., saying that whose
being so would bear in particular ways on how those further projects are to be
executed. If such is on the agenda, and if the representer succeeds in representing
things accordingly, he will make the contribution. Conversely, if such is (recog-
nizably) on the agenda, and if he can be understood as having represented things
accordingly—saying what, if so, would have that bearing—then such is reason so
to understand him. In what it would be so to understand him there is material
which could, if applied to identifying a way for him to have represented things,
effect selection.
I will elaborate this idea later. For the moment I merely illustrate. Guests are
coming. As Pia opens the wines to breathe, Sid sets the table. He has forks and
spoons in hand, but seems unable to find knives. Noticing this, Pia says, ‘There
are knives in the third drawer.’ Indeed there are. Suppose, though, that the third
drawer had contained Pia’s art supplies. These include a fair collection of matte
knives (roughly, handles mounting razor blades). One can understand there being
knives in a drawer so that the presence of matte knives counts as things so being.
But one might sometimes understand such talk such that such presence, on its
own, would not count as things being as thus represented. When I tell you where
the knives are, matte knives need not count as specimens of what I mean. The
words Pia used might be understood in either way. But she was to be understood
as speaking in aid of what Sid is doing—contributing in the way just scouted to his
project’s execution. Understand her in the first way and there is no such bearing.
Understand her in the second and there is. Such contributes at least to tracing a
path through the cloud of ways for things to be which contains all those which
are there being knives in that drawer on some understanding of there so being.
Whether this does, in fact, achieve selection for things being as Pia represented
them may remain an open question for the moment.
6.4. Agreement
rises from the grill. There is something Tara understood Sid to say. That is a psy-
chological fact. There is something Sid meant to say. That is another. There is then
the question what Sid did say, how he is, in fact, to be understood. That ‘to be’, like
the ‘to’ in ‘the thing to do’, removes us from the psychological. Our concern now is
with the logic of being as represented. Such a non-psychological question need not
have a determinate answer: there was an amusing misunderstanding, and there’s
an end on it. But it may. Perhaps Sid said what he meant to, and Tara misunder-
stood. Or perhaps the other way around.
What answers such a non-psychological question? What makes an answer
right? A starting point: for there to be allorepresenting is for it to be (made) recog-
nizable. How are we to understand this ‘recognizable’? Sid’s representing might not
be recognizable to a monolingual Latvian, nor to a Martian, or a cat. Such hardly
matters. His analyst might recognize what he intended. She might recognize this
of still more bizarre performances. Such again does not bear on how he ought to
be—is to be—understood to have represented things. There are, though, those
who ought to be able to understand him; those competent enough, and apprecia-
tive enough of his circumstances to do so. There may then be what they would
have a right to expect if then so addressed; how a competent understander who
knew what he should have of the circumstances would reasonably have taken Sid
to be representing things. So, the idea is, did Sid represent things being.
Who are these people? In the example, most Australians, one would suppose,
and some of the rest of us (most of us, if initiates in Aussie practice and patois). But
what matters is this. You and I (and most Australians, etc.) share a sense of what
to say in cases like Sid and Tara’s. It is a sense which indefinitely many other think-
ers—perhaps not all—either share, or could be brought to share through sufficient
familiarity with our ways of allorepresenting. Given the psychological facts—the
actual facts of our agreement in such matters—there is such a thing as what a com-
petent, appreciative audience for Sid’s words would be; such a way for a thing, or
group, to be as being such an audience. That there is is something we can recognize.
Given what it would be to be this, we can also recognize this to be a way for a thing
to be which is instanced. There are, further, recognizable facts as to how one who
was this way would understand Sid, and his representing, were he so addressed.
Such (non-psychological) facts would of course be recognizable to one who was
the way in question. They are recognizable to us because we are that way.
For any allorepresenting there is its audience—the sort of thing equipped
and placed to recognize it for what it is. To belong to Sid’s audience is to instance
the just-mentioned way for a thing to be. In other cases, it would be to instance
being competent and appreciative in re the representing there occurring. In any
case, the audience is, in principle, indefinitely extendible. Sid, as any Cartesian
thinker, shares with his audience those retrospective abilities I called a capacity
to be abashed. Such can be directed in concert at what Sid has done. Just this is
what allows for effecting the selection task for his allorepresenting. So it is with a
thinker’s allorepresenting. It should be stressed that this way of failing to be master
The Preserve of Thinkers 151
in one’s own house, so of relying on others for effecting a selection task, is reserved
for Cartesian thinkers. It works where a way the representer ought reasonably to be
taken is as doing what a Cartesian thinker might be doing (in the circumstances).
If the representer is not a Cartesian thinker, then, while mistaking it for one might
be understandable, he cannot have been to be taken as so performing.
Where allorepresenting has an audience (present sense), where there is such
a thing as what it would be to belong to it, to belong to the audience is to have
sufficient insight into how the representing is to be understood, and to have such
insight is to belong to it. When it comes to cases the audience is the measure of
what insight is here. That there is an audience may be manifest in its (extendible)
agreement. Just what is the audience’s role? Following Frege we may take it as in-
trinsic to any given way for things to be to reach just as it does. So fix a way and
nothing extrinsic to it, so no audience can make it reach in one way rather than
another. But what way a given way for things to be is is one question. What way is
such that Sid represented things as that way is another, as is what way one speaks
of where he speaks of lamb being on the barbie. If an audience provides no answer
to the first sort of question, it does provide the answer to the second. For a way for
things to be to be the way Sid represented things being (in speaking of his barbie)
is for it to reach to particular cases just as his audience (in the above sense) would
be prepared to recognize his representing (what he did) as reaching.
Any allorepresenting needs its audience, whether thinkers or not. For a Cartesian
thinker’s representing, the audience is of a certain sort. It shares a capacity: one, as
I put it, to be abashed. This capacity can be directed collectively at any instance of rel-
evant representing. It issues in acknowledgement of particular cases as thus reached
or not (not determined). Just here, in what a thinking audience would expect, the
crucial step is taken from the psychological to the nonpsychological—here from
holding forth to being true. What the audience would do. Where it is this audience,
is no longer a psychological generalization, nor a prediction. It is not like a statement
about what Sid, or Tara, or the average Australian would do. It is about how anyone
would respond to Sid if getting things right.
Our shared sensitivity to the conceptual performs this step for us. We achieve
selection tasks in ways which are the reasonable ones for the sort of representing
we engage in. Our sensitivity to the conceptual, such as it is, cannot be enlisted to
perform this step for an unthinker. If there are (parallels to) psychological gener-
alizations to be made about the unthinker’s doings, such need not be refractory to
us. If there are patterns in its responsiveness to the environment, we need not be
blind to these. But what it would thus do does not yet take us from the psycho-
logical to the logical, as something must if there is to be allorepresenting. Such a
step must be taken by the unthinker on its own, or anyway left on its own by us.
And the unthinker’s mere sensitivity to the presence of yellow, or pigs, in its sur-
roundings, whatever such may be, gives no right to construe any of its responses as
episodes of representing something as being yellow, or a pig, rather than as simply
detecting yellow’s, or porcine, presence. They effect no selection from within the
Against Strong Content
152
cloud in which that class of paths, cases of something being yellow are but an ele-
ment. The unthinker’s responses in its (presumably) hospitable environment give
no right to extrapolate from that subregion of particular cases to the space as a
whole. It would be anthropomorphism to construe its responding as it does to pigs
as, e.g., its telling us, or its peers, that a pig is about. If it were to present particular
cases as falling under generalities, the fact that we would be inclined to call what it
is doing detecting pigs, or yellow, gives no right to take those generalities to be at
all like those we can get in mind.
6.5. Deference
An unthinker could not take the step from the psychological (or mechanical) to
the logical in the way just sketched. The unthinker could not be to be recognized as
guiding execution of its representing as a Cartesian thinker would or might. Such
could not be the right thing to suppose of it. The Cartesian thinker’s way of tracing
a path through the space of particular cases could not be the unthinker’s. But per-
haps the unthinker need do no such thing. Perhaps he/it can simply contract the
work out, defer the selection task to some other source. One idea along these lines
would be: there might be a vehicle, identifiable independent of how it represents
things, which as such represents things as being some given way; and which, in
being the unthinker’s vehicle, would make it so that he/it so represented things.
Birds build nests but fail Descartes’ tests. So, Descartes thought, building
nests requires no intelligence. Who would think otherwise? Things are otherwise,
he thought, when it comes to holding conversations. If not all allorepresenting is
holding conversation, perhaps some, like nest building, is achievable by unthink-
ers. Something else would do the work for the unthinker that thinking does for us.
The above is one idea of what that something else might be: vehicles. These, the
idea is, would relieve the unthinker of the burden of selecting on its own. Its inca-
pacity would then not matter.
English sentences might seem a model for such vehicles. An English sentence
as such represents things as being a given way. It speaks as such of that way. The
sentence, ‘Monkeys fly’, speaks of what it does used or not, whether I take it to do
so, speak English, exist, or not. If it speaks of monkeys being flyers, then, where
I speak English, it would do so in my mouth. If it represents things as being a given
way then, the idea is, so do I in speaking it. The idea concludes: where I thus so rep-
resent things, for things to be as I represented them is just for monkeys to be flyers.
If English sentences so work, then, perhaps, so might other things, among which
things which would so work produced by (suitable) unthinkers. That English thus
models deference is an idea I hope now to dispose of. In its place I hope to put
this Fregean idea: the only way for a would-be content-bearer to come by the con-
tent of a thought, or of an element in one—to contribute to representings-as so
as to make them somehow about, say, relevant things being flyers—is for it so to
The Preserve of Thinkers 153
as Frege puts it, the necessary seriousness were missing, e.g., if he could not prop-
erly be taken so to have aimed.
English has a syntax. It thus generates an indefinitely large set of vehicles, its
sentences, from a smaller set of building blocks by fixed rules. What a sentence
says, or speak of, is then fixed by what its blocks do, plus the rules which structure
them in it. Some ideas for unthinking representers-as require these to have an
indefinitely large set of vehicles they might produce. If perceptual experience rep-
resented things as being given ways, for example, it would need to be able to repre-
sent things as any of indefinitely many different ones. So then it would need a stock
of vehicles built from a smaller set of blocks by some fixed rules. What a vehicle
said would thus be fixed by what its blocks contribute to this. Now, it may seem,
content may accrue to a vehicle merely by virtue of accruing anyway to its blocks.
But if Frege is right, this analogy breaks down. If a building block is to con-
tribute to representing-as, what has accrued to it anyway, independent of this
representing-as, then what has accrued to it anyway must be no less than the fea-
ture of representing-as. In representing-as, truth is made to turn in a particular
way on how things are. An element of such representing makes truth so turn, in
part, in that way. Being what so functions in the context of representing things as a
certain way—making truth turn, full stop, in that way on how things are—is what
a building block would need to be already to function in the imagined way as a
building block at all. Representing-as cannot emerge from mere syntactic struc-
turing. Combine what effect-represents the presence of something puce and what
effect-represents the presence of a Porsche however you like, and all you get so far
is something which effect-represents the presence of something puce and the pres-
ence of a Porsche. What a given vehicle would require for representing some given
thing as puce is, inter alia, a block which, in context, does that. If the block does
that in context by virtue of content it has anyway, then that block must already be
what functions to make truth turn, in part, on how things are. It is difficult to see
how any block could have come by this through interactions with the environment
which are any less than roles in representing it as being thus and so.
In 1882 Frege wrote,
I do not think that the formation of concepts can precede judgements, be-
cause this presupposes an autonomous existence of concepts, but I think con-
cepts arise through the decomposition of a judgeable content. (1882b, p. 118)
Concepts arise through decomposing whole thoughts. A thought is true of
things, where there is no question ‘Which?’ It is true of things, so true, tout court.
A (non-zero-place) concept is true of a thing. Truth-of, Frege notes, can be under-
stood only in terms of truth. For the concept (a thing) being puce to be true, say, of
Ed is for it to be true that Ed is puce. A concept (as here spoken of) is a common
feature in a range of thoughts—e.g., that Ed is puce, that Pia is puce, that that
torus is puce. . . . It is one way each reaches to particular cases. It fixes a generality
under which all such thoughts fall: making truth turn on what is puce. It just is a
The Preserve of Thinkers 155
common feature of those thoughts. There is no such feature unless there are such
thoughts. Concepts thus cannot precede thoughts.
So, too, for speaking of. There is no speaking of a thing as puce except in
the context of saying something as to what is or is not puce, or, more broadly,
representing things as some way the being which turns somehow or other on
things being or not puce. Speaking of a thing as puce (expressing the concept
of being puce) is not something which can precede speaking of things (catholic
reading) as thus and so. So speaking of a thing as being puce, expressing that
concept, is something a building block could do only in the context of its role
in the expression of whole thoughts. A building block might do that in isola-
tion only in that aspect of ‘speak of ’ in which to do so is to play a role in the
expressing of whole thoughts. Speaking of a certain way for a thing to be thus
cannot precede the expression of whole thoughts. So the accrual of content to
building blocks cannot precede the accrual of content to expressions of them.
What could not select a thought for a whole vehicle to express—a way for things
to be as how it represents things being—could not select a way for a thing to be
as what some building block contributes to such representing.
expresses (two different aspects of ‘express’ here) may be intrinsic to the ability to
author representing-as at all.
Words which aid recognition need not do so by virtue of their meanings; nor
at all. In a restaurant in Abbeville Pia asks for ‘ortalans.’ Of course one cannot order
ortalans, or not in this establishment, or in this salle, or season. But of course,
too, this is not what Pia (an Anglophone) meant. The way she is eyeing the our-
sins shows her to mean them. Habituated to tourists, the waiter simply brings Pia
her oursins. Pia managed to make recognizable what way for things to be she was
representing as wished for. She managed to request oursins. The sentence ‘I’d like
the ortalans’ speaks of a different way for things to be. But not every use of it to
allorepresent speaks of that way. Even where it does, that it does so need not be
enough to identify how it represents things as being.
In the restaurant, the waiter arrives with their plateaux de fruits de mer—
bulots for Sid, oursins for Pia. But he looks perplexed. Clearly he has forgotten his
orders. Seeing this, Sid tells the waiter that Pia ordered the oursins. Suppose Pia
garbled things, or Sid had done the talking. Are things as Sid thus represented
them? If Sid were reporting Pia’s progress with speech therapy, or with her patho-
logical shyness (say, to a worried mother), the answer might be ‘No’. But here Sid’s
words are in the service of a further project, placing orders. In contributing to such
an enterprise, one is hardly to be held responsible for who did the talking. One can
understand ordering oursins so that who did the talking does not matter. So here,
for reasons stated, one is so to understand Sid.
To hold Sid to have represented things in one way rather than another is to
hold him responsible for something—here something as to how things are. There
is a way things had better be if he is to be let off with discharging responsibili-
ties assumed. For what is he reasonably held to account here? Where should his
wishes be acceded to, aims honoured, where not? Is he accountable for who did
the talking; liable to praise or blame accordingly? Is such reasonably reckoned part
of the bargain in the liability he went in for in representing to the waiter as he did?
Above, I suppose, the answer is, ‘No’. He made clear what message he had to offer.
He need be, so is not to be, held responsible for more.
Responsibility gives a reading to that ‘ought to be taken’ in that step, in sec-
tion 6.4, from the psychological to the logical, from holding to being true, the
step in the ‘way one did allorepresent things being is the way one ought to have
been taken to have’. There is what Pia is reasonably held responsible for in then
lending Sid her Porsche (at least what one should have foreseen). There is the re-
sponsibility Sid undertook, signed on for, in saying Pia to have ordered oursins.
Allorepresenting is among a thinker’s means for undertaking responsibility. There
is then what it is fair to hold him to have signed on for in using those means then.
How he represented things as being is fixed thereby.
Sid makes recognizable two things about his allorepresenting. First, he is to be
taken as representing things being as he does in representing them as being a cer-
tain way, namely, such that Pia ordered the oursins. In his execution of his project,
The Preserve of Thinkers 157
the words he used are assigned the task of making this recognizable. Second, he is
to be taken as representing things as being that way whose instancing would have
a certain bearing on the way to execute a certain further project—the perplexed
waiter’s. He is to be—or asks to be—assigned responsibility accordingly. Perhaps
he could not be doing both these things jointly. Such is one way for it not to be pos-
sible to take him as he asks to be. Perhaps one cannot understand ordering oursins
so that whether Pia spoke does not bear on this, or that whether she did thus bears
on what the waiter is to do. But suppose we can. Sid ought not to be held to be
taking on responsibility he makes recognizable that he is not signing on for. One
ought not so to rely on him. In which case, these two features of what Sid was to
be taken to be doing jointly identify what it would be for things to be that created
way, being as he thus represented things. It is that way which reaches to just those
cases in which the world is such as to bear as it was to be supposed to bear on what
the waiter was to do, where things so being is understandable (might count) as Pia
having ordered oursins. It matters not whether it also has another name.
Sid represents things as he does in speaking of them as a certain way there is
anyway for things to be: such that Pia ordered oursins. He speaks on a particular
understanding of her having done so. One way to picture this would be as filling-
in. That way he spoke of, Pia having ordered oursins, reaches as such in a certain
way. Some range of cases is thus reached. Some other range fails to be. Other cases
remain undetermined. What it is for Pia to have ordered oursins yields as such
no verdict where Sid alone spoke to the waiter. The particular understanding on
which Sid spoke fills in some undetermined cases: on it some of these are reached,
some fail to be reached, by that way he spoke of.
If this is how things are, one might get a further idea. If Sid’s work of repre-
senting fills in understanding of that way he speaks of, so that his representing
things as that way reaches differently than that way on its own, then, perhaps, on
some occasion his representing simply fails to accomplish any such work. Then
things being as he represented them would reach exactly as things being such
that Pia ordered oursins does on its own. Perhaps an unthinking allorepresenter
could represent like that. What Sid thus did contingently would just be, neces-
sarily, its lot.
In what sort of case would Sid have done no such work? One might
think: when Sid spoke to the waiter, his talk of Pia ordering had an agenda. Our
talk often has much less of one. Suppose Sid simply wrote a postcard to Ed back
home: ‘Wonderful dinner last night. I ordered bulots, Pia ordered oursins.’ Not
much there by way of further purpose to be served. But now, must Pia have done
the talking for things to be as Sid wrote? Nothing in his writing this gives one any
reason to suppose so. So if she did, things are as Sid said. If Sid spoke for her, things
still are. But this is a special understanding of Pia having ordered oursins. What it
is for her to have done so does not, on its own, decide whether she needed to do the
talking. What it would be for her to have ordered can be understood in either way.
A case where no filling in was done would be a rather special one. Perhaps we get
Against Strong Content
158
6.7. Force
Those ways we can represent things as being—so those ways we can take things
to have been thus represented (whether by us, or by any representer)—are
such that where we represent things as some such way, it might be any of many
things to be as thereby represented. The last section concerned the capacities
drawn on in such representing. Its idea can also be put in terms of force. For
Sid to have represented Pia as having ordered the oursins in the way he did
is for him to have assigned that way for things to be a certain status: as to
be counted as among the ways things are where its being instanced is under-
stood as it would be for certain purposes. Assigning status cannot be just more
representing-as. This, too, would await a status. To coin a term, it is doing one’s
representing-as with force—here, in assuming responsibility, vouching for the
status thus assigned.
As force is usually conceived it comes in a small range of varieties: assertive,
interrogative, imperative, optative and so on. Things change if, as per above, in
representing things as some given uncreated way there is for things to be, one
can represent them as any of indefinitely many different ways. Throughout one
would present that way in which he so represented things with a certain force. But
it would need to be a different force in each case. He would assign that way for
things to be a status in re being among the ways things are. But that status would
be, not being a way things are full stop, but counting as a way things are when you
understand things so being in a particular way (in the last section’s image, with
a particular permissible filling-in). Force would vary here—even in an assertive,
or an imperative, or etc., case—according to the responsibilities signed on for, as
identified, e.g., as per the last section, in terms of projects to be taken as contrib-
uted to.
There is thus a selection task for force paralleling that for what way things
were represented being. How Sid represented things being in saying Pia to have
ordered oursins is fixed, not just by this being the way he spoke of, but also by with
what force this was presented—how it was presented as counting as a way things
are. An unthinker is as little equipped to effect the task for force on its own account
as it is the selection task for ways for things to be.
An unthinker would be overcome by allorepresenting, as a Tourette’s victim is
overcome with blurtings. The expletives are not the sufferer’s. The allorepresenting,
one might well think, would no more be the unthinker’s. Is there some default force
such unthinking representing might have? Perhaps it is a ‘purely generic’ asser-
tive (or imperative, or optative) force. ‘Purely generic’ here would be abstracting
from all particularities of ways of presenting that way for things to be in which the
unthinker represented things as it did. The status assigned would be: a way things
are no matter how you understand things so being. The usual way of thinking of such
abstraction is in terms of universal quantification: the way things are on all under-
standings of things being it.
Against Strong Content
160
The idea here is familiar in philosophy (cf. Clarke 1972). To paraphrase Clarke,
to see whether Pia really ordered the oursins, we stand back from any mundane,
local concerns such as how to tell the waiter what to do, and, purely considering
the concept ordering oursins as such, and the world as such, ask whether Pia’s
doings do, or do not, fall under that concept. Whether there is such a project of
pure inquiry is controversial. As indicated in the last section, if there is, it would
not be one reached by abstracting from all the particular varieties of assertive force
we have now seen there to be (all the ways of counting a way for things to be as a
way things are). It would just be another particular way of so counting things; one
to be applied where very special ends were to be served. And, as we saw in sec-
tion 6.3, classifying things according to the reach of some way for things as such
anyway draws on the full resources of a Cartesian thinker. So such abstraction, if
possible at all, does not relieve the unthinker of a burden. Nor is the burden one
the unthinker would have the capacity to discharge.
Perhaps, then, the task of force-selection is performed for him/it. We already
saw one idea for this: deference to vehicles. We saw already that that idea cannot
work. So perhaps the work is done by whatever thrusts allorepresenting on our
unthinker in saddling him/it with (producing or being) some vehicle. Putting
things in terms of force, though, brings out a point of Frege’s. It is that no vehicle
as such can impose a force on any representing (a version of the point above that
what gives force to representing-as cannot be just more representing-as). To put
the point one way, any way for an instance of ‘Pia ordered oursins’ to fail to be an
assertion is a way for ‘It’s true that Pia ordered oursins’ to fail to be one. When
assertive force is absent, ‘It’s true’ will not restore it, nor will ‘I assert that’, nor
any other form of words. Mutatis mutandis for any other force. In another ver-
sion Frege tells us that there is no assertive force “when the required seriousness
is missing”. Seriousness is not conferred by a vehicle. What is in question is the
seriousness with which it is produced. A thing cannot produce the necessary ser-
iousness by having representing thrust on it. It is the (would-be) representer who
(which) must be serious. This, as we have seen, the unthinker cannot be on its own.
This idea of abstraction, and of resulting generic forces, thus leads nowhere.
Might the unthinker then, perhaps, represent things as ways they are or not
(though not itself thereby representing truly or falsely), while doing so with no
force? When might a vehicle be produced forcelessly? A rhythm poet, or dadaist,
might produce English sentences simply for their sound—the sentence ‘Red balls
roll’, say, simply for the way it rolls off the tongue. Or a graffiti artist might spray
such a sentence on garage doors for its elegant shape. Most red balls probably do
roll, if not made of glue. Such is not what (if anything) makes the poet right. If his
interest in sound is pure enough, then while he wrote a sentence which speaks of
a way for things to be, he did not thereby engage in any representing-as at all. And
if, as we stare at the garage door admiringly, a red ball rolls by, well, what a coinci-
dence! But it is just a coincidence. As some philosophers have it, in (a) perceptual
experience the world is represented to us as a certain way. If we see a pig under an
The Preserve of Thinkers 161
oak, say, then perhaps as such that a pig is beneath an oak. But if this representing is
conceived as forceless, then it might equally well represent things as any other way,
say, as such that cool waters run deep, or Pia drives a Porsche. Experience’s so rep-
resenting things may mean (factively), effect-represent, or indicate, or make likely,
that a pig is beneath an oak. But if the representing here is forceless, then it is not
through its content that such meaning is effected. It is not as if a reason thus created
for thinking a pig beneath an oak might be that experience, or this representing in
it, might be right. Whether it is right or wrong cannot matter here: without force
there is no way for it to be either. Representing-as thus cancels out. Representing
Porsches as fast would do as well as representing a pig as beneath an oak for nature’s
signal that a pig is beneath an oak. For what meant in this sense of meaning, to
represent things as the way it means they are would just be a curious accident.
Force is part and parcel of the step by which we move from the psychological to the
logical, from mere effect-representing, or its relatives, to that three-place relation,
representing-as.
There is finding, or marking, instancings, or the instancings, of some given way for
things to be; and there is presenting things as some given such way. The one thing
is not the other. But some might hope for the second to emerge out of the first. This
section explores that idea.
An unthinker lacks capacities which, so far, appear essential for allorepre-
senting; any capacity which might permit that leap from the psychological (or
its counterparts) to the logical, from the psychology of holding forth, to the logic
of being true, which allorepresenting is per se. How, then, might unthinking
representing-as ever be thought a possibility? One prominent idea is that allorep-
resenting might emerge out of (the maintaining of) patterns of effect-representing,
aided, perhaps, by the point of maintaining them. The allorepresenting would be
by, or in, that in which such patterns were maintained. This section explores that
idea.
The simplest patterns are generalizations, the simplest generalizations univer-
sal. Those empty seats in the theatre (or their presence) effect-represent casting Sid
in the lead role. Empty seats in a theatre might always do that, if Sid got around
enough. Or, to complicate things, they might usually, or normally, or (other modi-
fier) do so. So far, it is the presence of those empty seats which does the represent-
ing. If allorepresenting emerged here, what would do it? Frege writes,
No one can be prohibited from adopting any arbitrarily occurring event or
object as a sign for whatever. (1892, p. 26)
Empty seats could be appropriated as a sign that Sid plays the lead. They would
thus be a vehicle of representing-as. What would do the appropriating in the
Against Strong Content
162
be: such that Sid was in the lead. If they did that, they ipso facto effect-represented
the instancings of countless other ways for things to be—countless co-denizens
of clouds within the conceptual that Sid being in the lead inhabits. Such would
remain so if empty seats always effect-represented Sid in the lead—wherever, that
is, nature is so arranged. Effect-representing does not perform the selection task.
Nor would generalizations of the kind just scouted.
But if allorepresenting has not yet emerged, perhaps we are looking at too
simple patterns. Here is another idea. Sometimes we can say: A effect-represents
(or would effect-represent) B if all is/were going right. That needle on the gauge
effect-represents the tank’s being half full if all is going right. Those hands on the
dial (or their present position) effect-represents its being 10 o’clock local time if
all is going right (if Sid remembered to reset his watch). Such patterns, if any,
are those from which someone might think allorepresenting could emerge. I will
expand that idea.
There are designs which, if realized, would make for A effect-representing
B. A device, or mechanism, or system, or (perhaps) phenomenon may realize such
a design. Man, or nature, may provide such devices (or etc.). (Examples above.)
Sometimes by this design, sometimes, perhaps not, the device produces, on occa-
sion, a certain outcome (or type of outcome), A. The output may be a product—a
signal, say, or an effect—or it may be the device’s going into a certain state, A (or
of type A). When such a device is working as per design (as any device is liable, on
occasion, not to do), A, as thus produced, effect-represents B (what it was designed
to). A would occur only if there were B to thank. Such a device may be for real-
izing this design in this sense: but for some need thus to connect A and B, the
device would not have been created. I will call such a device a B-detector: where
the device produces A as outcome, you can bet on B’s occurrence (if all went as per
design). Bs are historical occurrences, e.g., of porcine presence. What the device
detects is thus particular instancings of things being a certain way, e.g., such that
a pig is present.
Depending on what it was detecting, a detector might need to exploit com-
positionality. If, say, it were detecting where animate things were, it would need
indefinitely many dedicated outcomes for each of indefinitely many arrays of loca-
tions relative to it at which there might be such things for it to detect. Composing
outcomes would be called for. For present purpose such changes nothing.
A design might also be for (as I will speak) locating (cases of) B: not just for
making it so that A, if it occurred, would effect-represent B, but also for making it so
that, all working as per design, B, if suitably occurring would be effect-represented
by A. So if B is (suitably) occurring, you can bet on A, again all going as per design.
Now the hope is, either for detectors, or for locators, that, in detecting, or locating,
as the case may be, they will also be representing things as being that way whose
instances are thus detected or located.
On one notion of device, a device for detecting (or locating) would work in the
first instance by responsiveness to proximal provocation. It would effect-represent
Against Strong Content
164
such provocation. If allorepresenting were to emerge from its workings, one might
see it as simply representing things as being such that there is such provocation—
in this sense simply representing the proximal. But in most cases of interest, the
device will be for detecting, or locating, the distal, e.g., porcine presence. When it
works as per design, it will effect-represent those distal things too. And the hope
will be that those ways for things to be whose instancings it detects or locates will
be the ways it allorepresents things being. What is thus responsive to the distal
in its responsiveness to the proximal is inherently subject to what would be for it
ringers: there might be the right proximal provocation without the wanted distal
happenings, and vice-versa.
Our samples, so far, are manmade. But such devices can be natural. Pigs chew
straw, it is said in some parts, when it is about to rain. In perceiving, it is sometimes
said, we experience the world being represented (to us) as being thus and so. What
of this case? Pia sees the pig before her. Her doing so is thanks, inter alia, to there
being one. So it effect-represents there being one. Thus far all is in place for her to
be a pig detector, her seeing a pig being the outcome which would be detection.
But it takes no design to realize this connection, a fortiori not one which works via
responsiveness to the proximal. There is no such thing as her seeing a pig failing to
effect-represent presence of that pig. So, though I have tried to remain ecumeni-
cal on the crucial point here (not to anticipate routes by which someone might see
representing-as as emerging), Pia’s case to be a detector by virtue of her capacity to
see pigs sits ill with our present notion of a detector—one designed to fit the intu-
ition that nature might make representing-as emerge.
Design comes into the picture here when it comes to locating cases of porcine
presence. What needs explaining is not that when Pia sees a pig there is a pig, but
rather that she does such a thing at all. Indeed, Pia (if adequately sighted) is such
that, by design, when a pig is before her (and she is looking), all going well, she will
see it. Such is no thanks to her responses to proximal provocation (though it may
depend on some processor’s responses). Nor, correlatively, is this capacity exposed
to ringers: the indicated outcome with no pig before her. Nor to a ringer for no pig
(hence a pig) which she experiences visually without thereby seeing a pig.
If Pia sees a pig, then, though this does effect-represent porcine presence,
such does not bear on the reason she thus gains for taking there to be a pig before
her. Her reason gained is not given by the fact of her seeing a pig, but rather by
what she sees—a pig before her—which she can recognize as a case of a pig being
before her (as being such that a pig is before her). As Frege notes, such recogni-
tion is a function of thought, not vision. What is gained is reason so to suppose—
nothing short of proof that there is a pig—which need not consist in reasons so
to suppose. Such reason gained, considerations of effect-representing could add
nothing, could not so much as be reason so to suppose. If representing-as emerged
from this effect-representing, it would offer exactly that much.
Things would be different if for Pia to see the pig is for her to be in, not just
that state, but some other visual state—one, say, of being appeared to thus—which
The Preserve of Thinkers 165
she could conceivably be in even were there no pig. Then it needs explaining how
this state in fact effect-represents a pig before her (when it does). A design is
called for; one whose realizing would maintain the right relation (all going well).
A design for given proximal responsiveness seems indicated. With this state as
the indicated outcome Pia fits the model of a pig detector. If allorepresenting
ever does emerge out of effect-representing conditions might now be ripe for
this. Such allorepresenting might even have a point. It might give Pia reason to
suppose of what this state is visual awareness of that this time it effect-represents
what it ought, porcine presence (though it is hard to see how this could ever be
good reason). Reason the more, I suggest, to find this a bad picture of perception.
For present purposes, though, I simply mark this as the picture in which the idea
of representing-as in perceptual experience might look promising.
Back to the general question how representing-as might emerge. What I now
want to stress is the distance by which detecting instancings of some way for things
to be, or locating its instancings, falls short of presenting things as being that way,
or placing things under given generalities—as one does in representing-as. Such
will bear particularly on the second problem above. A good way into the matter
is the following. A detector (or locator) works according to a particular design;
a design for detecting whatever it is that it detects. It works via the proximal,
and it works in a particular way. A pig detector, e.g., may be sensitive in its out-
comes to a pig’s distinctive snout: whether it produces that outcome which is to
effect-represent porcine presence depends on whether it has detected (or done
what, for it, ought to be detecting) such a snout. And it detects such snouts by
marks of such which, by design, would be proximally accessible to it, once again,
all going well. But by what marks, or means, such things as pigs or Porsches, yel-
low or snores, are detectable depends on the environment. Equally for anything
else liable to be detectably present or not in the sublunary world we inhabit. So a
condition for a detector, or locator, so much as being that is that it work in an en-
vironment hospitable to its ways. This point entered the picture already, as we saw,
with the very idea of proximality.
You can tell a pig by its snout. You can rely on this given how things are. But
no one supposes that to be a pig just is to have such a snout. Plastic surgery alone
rules that out. To be a pig is not just, and not per se, to look some particular way.
So in the wrong environment (e.g., too much cosmetic surgery), a pig-detector
which worked by means of snouts simply would not be a pig detector. We can rec-
ognize when it would not be. We can look at its workings, on the one hand, and,
on the other, at the reasons for and against counting what it would be identifying
as pigs (if detecting at all) as pigs. We can exercise our capacities for retrospection,
capacities to be abashed. An unthinking detector has no such capacities, cannot
be abashed.
The unthinking detector lacks a capacity we have: sensitivity, case by case,
to what bears, or might, and how it might, on whether that case is to be counted
as one of instancing any given generality—notably, here, that generality whose
Against Strong Content
166
instancings are to be detected. As noted long ago, such capacities are reserved for
Cartesian thinkers. An unthinking detector thus could not be sensitive to what
distinguishes this way for things to be, such that there is a pig beneath the oak,
from countless others, notably, others inhabiting clouds around it. This does not
matter to detection, but certainly does to presenting. For presenting things as
being such-and-such way, a selection task must be achieved. That way things are
presented as being must be distinguished, in the presentation, from its fellows
within conceptual, notably those overlapping with it through some proper part
of the space of particular cases, but diverging from it in other proper parts. Such
selecting is beyond the powers of an unthinking detector (or locator). Mere detec-
tion does not demand it. Assume the detector/locator in a hospitable environment
where he/it always gets things right. If it is representing things as some way it is
detecting in this environment, such fixes something as to how that way reaches. It
reaches to these cases. But such is only a proper part of its reach. The unthinking
detector/locator has no capacity to see when it would have left a hospitable envi-
ronment or how, or, when it has, what detection then would be. It thus cannot do
what any thinking allorepresenter can. Which robs us to our right to suppose that
the notion environment hospitable to its detecting has a determinate sense or appli-
cation, that it is so much as fixed what a hospitable environment for it would be.
A detector might fail either through migration into hostile territory, or thanks
to some one-off ringer. If he/it represented things as those ways whose cases he/it
was thus detecting, such would be cases of representing falsely. But the unthinking
representer cannot represent itself as in a hospitable environment. It is blind to when
this would be so. The Porsche detector which blinks each time a Porsche passes, put
in the world of knockoff Porsches, can do no more than carry on. Its detection work
in no way equips it to approach the question whether it is in a knockoff world. That
it is not cannot be how things are according to it. Nor can it be held responsible,
where ringers are about, for whether what it blinks to is a Porsche. Such thus cannot
really be, in such circumstances, how it represents things being. If it represented-as
at all, it might just as well be that those instancings to which it continues blinking
are just those of the way it represents things being.
Detection buys the unthinker no standing in the realm of allorepresenters,
since it does not equip him/it to effect selection. With which the idea of allo-
representing emerging out of detecting, or locating, collapses. What qualifies an
unthinker as a detector is not what could qualify it as a presenter, or placer, so as
allorepresenting-as (as presenting something as falling under some generality).
Allorepresenting cannot thus emerge.
6.9. Collapse
A pattern which made for detecting (or locating) instancings of some bit of the
conceptual would not thereby make for presenting anything as falling under any
The Preserve of Thinkers 167
So things stand with the second of our two problems. I return now to the
first. I approach it first through this question. If allorepresenting arose out of
some pattern of effect-representing, with what force would this be done? An
unthinker does no autorepresenting (I here bracket dogs and cats). Where there is
allorepresenting, there is what makes it recognizable, its vehicle. Here this is to be
some occurrence—some production or its product—which instantiates the pat-
tern, is produced by a design for maintaining it. It is to be, in present terms, what
is reserved by design in some detector for signalling detecting of some instancing
of the way as which things are thus represented. The problem I will scout now
arises from the fact that, to speak a bit loosely, all the information carried by the
(supposed) representing is carried already by its vehicle.
The occurrence of this vehicle, like any occurrence, effect-represents all that
to which it owes thanks. If all went well—if it was produced by design, and the
environment was, even locally, hospitable (no ringers)—then this occurrence
effect-represents what the design is a design to detect. On a given understanding of
going well, hospitable, ringer, and so on, this would be the instancing of some way
for things to be. So the occurrence of the vehicle is liable to give reason to think
that there was such an instancing. It gives precisely as much reason to think this
as there is then reason to think that all went well, surroundings were hospitable,
and so on—conclusive reason, plain proof, where such things are not in doubt. The
vehicle itself (properly, its occurrence then) has that much significance. There is,
anyway, this to be recognized as to what to think and do. So much is recognizable
to one blind to the fact that the vehicle is a vehicle of allorepresenting; to what
allorepresenting there thus was, or to there having been any, so long as he recog-
nizes the vehicle as produced, as it was, by what incorporates such a design—by
what would, if working well, etc., thus maintain that pattern of effect-representing
which it is a design for maintaining.
Now suppose allorepresenting to have emerged in this operation of what pro-
duced the vehicle (the detector). What reason does this allorepresenting give for
thinking and doing? How does it, or its occurrence, bear on what the thing to
think, or do, would be? If it emerged from the pattern, as per above, then for it
to have occurred is for it (or its vehicle) to have been produced in the maintain-
ing, by design, of the relevant pattern—one such that, things going well, etc., its
(or its vehicle’s) production would effect-represent the instancing of what is thus
effect-represented. It would have occurred on just that condition on the vehicle’s
effect-representing such instancing. To recognize it as the representing it is is to
recognize it as just such a designed production. So its occurrence, if recognized for
what it is, gives just the reason to think things the way it represents things being as
the vehicle’s occurrence itself gives for thinking this. And if it really thus emerges
from the pattern, its mere occurrence can give no more.
Whereas it is essential to allorepresenting that it (or its occurrence) can bear
on the thing to think (or do), give reason to think things one way or another
(where they are represented to be F, that way) which there would not be anyway,
Against Strong Content
170
messages. From their occurrence one may (sometimes) conclude that there is a
Porsche, as from the pawings in the dust on the trail ahead one may conclude that
the wild boar are in rut.
The difficulty with emergent allorepresenting has been, so far, that what is to
make this representing recognizable as what it is is identical with that in it which
gives reason for supposing things as to how things are, or are to be. Whereas allo-
representing is always a source of new and distinctive sorts of bearing on questions
what to think or do. It is a source of reasons to think things that an unthinker could
not give. In spelling this out a bit, I can reinforce and deepen the points just made.
Suppose that Sid tells us that a Porsche is in the drive. His doing so might,
by any of countless routes, effect-represent the presence of a Porsche in the drive.
(For example, perhaps in his seeing Pia chatting with the countess, by the French
doors leading to the garden.) It could, but need not, effect-represent his ability to
tell a Porsche at sight. Such things may be nice if so. But they are not what make
his representing recognizable. No pattern of effect-representing makes for his rep-
resenting, nor for it being recognizable. What makes for his representing, so for its
recognition, is independent of any patterns of effect-representing in which such
performance may stand (Chomsky’s point). Which makes room for Sid to achieve
recognition through choosing means according to his insight into that audience
(to which he belongs) for which (in the sense of section 6.4) he allorepresents; ac-
cording to that audience’s shared sense for how such means, deployed then, would
be to be taken.
What makes Sid’s representing recognizable is thus also independent of that
which gives it the bearing it does on what to think and do, which makes it the
reason it is for thinking this or that. If Sid represented a Porsche as in the drive,
there would be a Porsche in the drive, provided that he was then executing a proj-
ect of seeing how things were, knowing how to do so and when he would be. Where
this condition is recognizably met, his so representing things gives conclusive
reason to take a Porsche to be in the drive. More generally (in parallel with that
supposed emergent representing-as by a detector) Sid’s saying so gives as much
reason to think so as there is reason to think the condition met. Notoriously we
do not always tell the truth; nor are always able to tell whether we are doing so
or not. But being serious (one’s project being one of saying how things are) and
knowing what one is talking about is something we, sometimes, can manifestly do.
So, sometimes, we can recognize the reason Sid’s saying so gives to take there to
be a Porsche in the drive in recognizing the reason there is to take him to be thus
engaged. Sid’s saying so then gives us such reason.
For Sid’s representing to give the reason it thus does for thinking a Porsche in
the drive, it need participate in no fixed, specifiable, pattern of effect-representing
(a pattern maintained in Sid in some given way). Nor is it in recognizing such a
pattern that such reason is recognizable to us as given. We need know nothing of
such things. Nor could such reason emerge from such participation. So nor could
it be such participation which made it recognizable to us as given. Thus is the
Against Strong Content
172
are not thoroughly without effects, but their effectiveness is of a wholly other
sort than that of things. . . . their effects are triggered by the doings of think-
ers. . . . (cf. 1918, p. 77)
Instantiating, thinkers’ doings are what it takes to place thoughts in allorepre-
senting. Instantiating again, perceptual experience is nothing like being repre-
sented to.
6.10. Afterword
The question this volume poses is whether perception has content. I have under-
stood content as the content of representing-as (representing something as some-
thing), and then that for some representing-as to have the content it does is just for
it to represent things as being as it does. I have then tried to answer the volume’s
question by answering a broader one: Where might there be representing-as at all?
The answer I have argued for is: Only a thinker (and that by which he executes his
projects of representing) can engage in representing-as.
How do I suppose this to answer the volume’s question? First, I suppose
perceiving to be a passion in this traditional sense: it is something we undergo;
something imposed, inflicted on us, something we suffer (like the party next
door at 3 a.m.). Accordingly, I suppose that if there were content in perception,
it, too, would be part of what is inflicted on us, or integral to its infliction. The
point of this is just that it would not be the content of our representing-as by way
of responding to what we see—e.g., responding to the monkey with the tin cup
before one by taking there to be a monkey before him. And, if the perceiver’s own
candidature for the role of representer is thus scratched, no other thinkers appear
in the vicinity to do the needed work. That perceiving (or even, more broadly,
perceptually experiencing) is a passion in this sense has been challenged. I will
return to this. Anyway, such is the strategy.
If I am right, then seeing perception as with content is something like seeing
one’s teacup as (literally) having it in for him. Why, then, should this idea about
perception be as immensely appealing as it seems currently to be? One short answer
would be sociological (or epidemiological): things go around. A short philosophical
answer would be: inattention to details, or, more generously, a congeries of gram-
matical illusions. I except from such answers one variant of representationalism on
which a positive answer is portrayed as the only solution to a certain how-possible
question (in the spirit of Kant). For the moment, though, I bracket this exception.
A first question to ask, then, is: Just where is content supposed to be located in
perception? There are, I think, three main answers: first, in the objects of percep-
tion (or of perceptual, e.g., visual awareness); second, not there, but as otherwise
presented to us in our perceiving (the inflicting on us of the visual awareness thus
enjoyed); third, in certain information-processing mechanisms or their states. On
Against Strong Content
174
the first two answers, awareness of the relevant representing, or at least affordance
thereof, is part of the bargain in perceiving. So a perceiver should be prepared to
recognize (perhaps with the help of a bit of elicitation) just how, in his experience,
things were represented as being (or at least when things are, when not, as so rep-
resented). This idea is mandatory insofar as the representing here is supposed to
be a source of knowledge, or, again, deception. On the third answer the perceiver
has no, or no direct, access to the relevant representing; if he has access at all, it is
inferential.
Putting last things first, motivation here is often in part reductionist. The core
idea: for something to look, e.g., yellow to me is for something intracranial to be as
it would be if registering something’s actually looking yellow—thus to be in a cer-
tain representational state. It is anyway implausible that such a state would be one
of representing the world as being thus and so (e.g., something in it as being yel-
low). The comparison here would be with a digital camera. Perhaps a file contains
instructions for recreating the image recorded on it. But the camera merely regis-
ters, does not represent. Nor does either it or the file represent the world around
it as being thus and so. In any case, the answer to my more general question rules
out such locations for representing-as.
As for objects of perception, it was, perhaps, easier to take them as engaged
in representing-as in the days when one could, with a straight face, maintain that
those objects (e.g., of vision) did not cohabit our world with us—not any such
thing as a champagne flute or the bubbles therein. Objects of perception literally
out of this world might invite a certain Lockean ploy. Few would hold, though,
that that penguin on the rock, or teacup on the table, represents anything as being
anything. So, with the demise of the view which invites such a ploy, it becomes im-
plausible that the content of perception is so located. (Though, disclaimers to the
contrary, reports of the demise of this view may be premature.)
The remaining view, then, is that while the representing-as in perception is
not itself an object of, e.g., visual awareness, in being visually aware of what we
are (in seeing what we do) we are also, ineluctably, aware of things (notably, in
the visual case, the scene before our eyes) being represented as being thus and
so (where it is not ourselves who are doing so). Generally speaking (perhaps not
always) the content of the representing here is thought of as, so to speak, indexed
by the way things look to the perceiver. For example, it is not going to be that,
while the sky looks blue to me, my experience represents it as being red.
If representing-as could not diverge from looks in such ways, such would be
a sign of mere idling—unless the point is to answer a certain how-possible ques-
tion: How could perception make the world bear, rationally, for the perceiver on
what to think? How could it make us knowledgeable? How could we ever, e.g.,
know that that monkey has a tin cup because we can see the cup? The obstacle
here is a doctrine. It concerns rational relations (relations by which things can
bear on what is true). The idea is: only representations can stand on either side of
a rational relation. For, it might be said, rational relations come on the scene only
The Preserve of Thinkers 175
with truth thus properly in the picture. For familiar reasons (and more to come),
it is implausible that representation could really dispel such a how-possible worry
(were there one). In any case, the right way with the worry is to jettison the doc-
trine—for doing which Frege provides compelling reasons. (How could there be
any truths unless some things not fit for truth could bear on whether what could
be true was so?)
A last idea: perhaps seeing a monkey with a bellman’s cap and tin cup is not a
passion in my initial sense. Thinking that one faces a monkey is, per se, representing
things to oneself as such that such is what one faces. Similarly, the idea is, seeing the
monkey standing is, per se, representing things to oneself as some given way. But
there is less than no reason to think this. As Frege notes, ‘see’ in ‘see that’ is not a verb
of perception. The monkey is (now) on the table, too near my glass. That there is a
monkey there is neither on the table, nor in any other location. It is hence, as Frege
puts it, not something which can form images on retinas. Seeing as is sometimes
perceptual, sometimes not. (A locution with many uses.) But in any case seeing a
monkey as a monkey is certainly not something done routinely, or merely in seeing
a monkey. There is no reason to think that any such thing is intrinsic to perception.
It is easy enough sometimes to see things where they are not. It comes easily to
many to see representing where it is not, namely, in perception. The prophylactic
against this is to mind one’s grammar. This, though, is only to be done case by case.
I hoped, in my present contribution, that some general considerations about what
representing-as is might still that urge which calls for prophylactic.
Finally, what is the importance of the question whether perception has (rep-
resentational) content? Not least in that its answer bears on a project Frege laid
out clearly: understanding those two utterly different main forms of awareness we
enjoy of how things are. Perceptual awareness (that awareness which perceiving
is per se) is one. Awareness in thought—e.g., appreciating, realizing, recognizing,
seeing, noticing that such-and-such is so—is another. What one could be aware of
in the one of these is, as Frege stressed, what one could not be aware of in the other.
One can see the meat fall before one onto the white rug. One can watch its fall.
The meat is on the carpet. By contrast, that the meat fell is neither on, nor under,
it. That the meat fell is unlocatable. Correlatively, it can form no images on retinas.
So, though looking at the meat on the rug, one may say, ‘I see that the meat fell
out of the grocery bag onto the rug’, here ‘see’ is a verb, not of perception, but of
cognition. (Thus, that one need not be, and presumably was not (observing the
rug), visually aware of all the events thus mentioned.) It is nonsense to speak of
being visually aware of ‘that the meat fell’. By contrast, there is something the meat
looked like as it fell onto the rug; something its fall looked like. But one cannot
think such bits of history as the meat’s fall. There is no ‘realizing that the meat’s
fall’. For history to have taken the turns it thus did and for its being as it then was
to be a case of a certain general way there is for things to be—such that meat had
fallen—are two quite different things. Undergoing the sight of falling meat is not
bringing anything under a generality.
Against Strong Content
176
which the objects of our perceptual awareness might come to bear for us on what
to think, such representings simply call for more of those same means (whatever
these might be) before they can have rational import for the perceiver on what to
think. For understanding the relation between our two forms of awareness, they
are merely Ptolemaic remedy.
Third remark. If, gazing at the rug, we are aware of things being such that
tonight’s T-bone is on it, and if, moreover, things are so represented in the repre-
senting then inflicted on us, this last part can be only incidental to the case. For
either this representing merely endorses what we can already recognize in seeing
what we thus do, the T-bone in plain sight, we knowing our meat, or we cannot
tell by looking whether it is T-bone on the rug, in which case it is not percep-
tion’s prerogative so to inform us, whether in some occurrence of representing or
otherwise.
Such awareness is not inflicted on us as our awareness of the presence of this
representing-as is. It is the awareness enjoyed in our taking it, knowledgeably, that
such-and-such is so (in our holding such a stance). (This should not be taken as
the discovery of a new, third, form of awareness.) Whether we are then aware
of things being any of the ways they were thus represented as being (assuming
they are any of these ways) depends on what we are aware of—notably in percep-
tual awareness—other than the presence, or occurrence, of that representing—of
what such other awareness positions us to judge knowledgeably. (If in that rep-
resenting the meat is represented as a T-bone, it is seeing the meat which allows
us to see whether this is so.) The excrescence on perceptual awareness that such
representing-as would be could only be at best annoying.
References
Chomsky, Noam (1959). Review of B. F. Skinner’s Verbal Behavior. Language, 35(1), 26–58.
Clarke, Thompson (1972). The legacy of skepticism. Journal of Philosophy, 69(20) (November
9), 754–769.
Descartes, René [1637] (2000). Discours de la méthode. Paris: Flammarion.
Diamond, Cora (1991). Frege against Fuzz. In id., The Realistic Spirit (pp. 145–178).
Cambridge, MA: MIT Press.
Frege, Gottlob (1882a). 17 Kernsätze zur Logik. In Nachgelassene Schriften (pp. 189–190).
Hamburg: Felix Meiner Verlag, 1983.
Frege, Gottlob (1882b). Letter to Anton Marty. In Gottlob Frege’s Briefwechsel (pp. 117–119).
Hamburg: Felix Meiner Verlag, 1980.
Frege, Gottlob (1892). Ueber Sinn und Bedeutung. Zeitschrift für Philosophie und philoso-
phische Kritik, NF 100, 25–50.
Frege, Gottlob (1892–95). Ausfuhrungen über Sinn und Bedeutung. Nachgelassene
Schriften, 128–136.
Frege, Gottlob (1918). Der Gedanke. Beiträge zur Philosophie des deutschen Idealismus, 1
(2), 58–77.
Against Strong Content
178
McDowell, John (1994). Mind and World. Cambridge, MA: Harvard University Press.
Kaplan, David (1989). Demonstratives. In J. Almog et al. (Eds.), Themes from Kaplan (pp.
481–563). Oxford: Oxford University Press.
Putnam, Hilary (1988). Representation and Reality. Cambridge, MA: Harvard University
Press.
Waismann, Friedrich (1979). Ludwig Wittgenstein and the Vienna Circle. Oxford: Basil
Blackwell.
Wittgenstein, L. (1922). Tractatus Logico-Philosophicus. London: Routledge and Kegan Paul.
7
As most readers of this volume will be aware, the dispute between disjunctivism
and the common kind theory in the philosophy of mind centers on the metaphysics
of certain types of conscious events: veridical perceptions, illusions, and hallucina-
tions. Traditionally, these three types of events—sensory experiences—have been
taken to be of the same kind, in the strong sense of sharing the same fundamental
nature. What binds them together is said to be (for example) their intrinsic qual-
itative character, or their intentional content, or a certain adverbial analysis. In
contrast, the disjunctivist I will be discussing holds that veridical perceptions are
essentially different in kind from illusions and hallucinations.1 He thinks that the
mind-independent entities that are the intentional objects of our experiences are
constituents of the experiences that are veridical; and for obvious reasons, no such
constituency can obtain in the case of illusions or hallucinations.2 (Disjunctivism
is sometimes called a “relational” theory of perception, because of this constitutive
relation between perception and its objects.)
The disjunctivist acknowledges that veridical perceptions, illusions, and hal-
lucinations are all experiences, but he claims that what unites them as such is just
their indiscriminability from veridical perceptions—not their qualitative or inten-
tional or adverbial properties. Specifically, a mental event is an experience as of Φ
just in case it is indiscriminable from a veridical perception of Φ. The notion of
indiscriminability at work here, following Williamson (e.g., 1990), is meant to be
epistemological: experiences are indiscriminable if they cannot be told apart, i.e.,
cannot be known to be distinct, through introspection. M. G. F. Martin explains:
We need not look for some further characteristics in virtue of which an event
counts as [e.g.] an experience of a street scene, but rather take something
to be such an experience simply in virtue of its being indiscriminable from
1
Here I ignore many significant differences among disjunctivist positions; see, e.g., Haddock and
Macpherson (2008) and Soteriou (2009) for detailed surveys.
2
Actually disjunctivists disagree about the proper treatment of illusions; I skate over the issue
here, but see again Soteriou (2009). 179
Against Strong Content
180
3
Some exceptions are Hardin (e.g., 1988), Burns (1995), Raffman (2000), and Graff (2001).
Disjunctivism, Discrimination, and Categorization 181
4
As Siegel (2004, p. 206, e.g.) points out, the alleged nontransitivity appears to pose comparable
problems for the common kind theory.
Against Strong Content
182
5
For present purposes I will suppose that a stimulus appears Φ to a subject at a time just in case
he judges it to be Φ upon inspection at that time; in particular, two stimuli appear the same in hue just
in case the subject judges them the same in hue upon inspection. Of course, in some contexts such a
supposition would be problematic, even question-begging, but it shouldn’t cause trouble here.
6
Discrimination and categorization are probably functions of distinct psychological mechanisms;
see, e.g., Pernet et al. (2004); Mirman et al. (2004).
Disjunctivism, Discrimination, and Categorization 183
perceivers can discriminate more perceptual values than they can identify on most
stimulus dimensions is a basic tenet of psychophysics. Perceptual memory has its
limits.) Similarly, a subject may be able to tell by introspection that her experiences
of those samples of red are of different types (shades), without being able to iden-
tify either of them as an experience of red28 or as an experience of red29.
As far as I can tell, when philosophers of perception talk about the alleged
nontransitivity of indiscriminability, they almost never intend the statistical rela-
tion. Martin and Siegel certainly don’t; they have in mind a phenomenal relation.
In calling two stimuli indiscriminable they seem to mean either that the stimuli
invariably appear the same when compared or that they appear the same in some
same/different comparison at some time. Since there are no stimuli that invariably
appear the same, we need to read Martin and Siegel in the second way, as referring
to an occurrent relation of sameness of appearance.7 We will suppose it’s the latter
relation they believe to be nontransitive.
Perhaps the most compelling reason to think that appearing the same is non-
transitive is the existence, or at least the possibility, of phenomenal continua, viz.,
apparently continuous progressions in hue (loudness, pitch, etc.). In light of the
distinction between indiscriminability and appearing the same, how should we
characterize phenomenal continua? They are not plausibly identified with indis-
criminability series, in which neighboring members do, but the endpoints do
not, appear the same in a certain percentage of comparisons. A phenomenal con-
tinuum is supposed to be phenomenal, hence apparent upon inspection. As a first
approximation, a phenomenal continuum is a series of simultaneously presented
stimuli in which neighboring items appear the same but the endpoints appear dif-
ferent, at a given time, to a perceiver who proceeds along the series giving each
pair of stimuli a good straight look (listen, taste, etc.).8 More precisely, a phenom-
enal continuum is a continuous progression in appearance that is instantiated, for
an observer at a time, by such a series. The phenomenology of phenomenal conti-
nua is baffling because, given perfectly constant viewing conditions, the first and
last items appear different and yet nowhere between the two is any local difference
in appearance discerned.
Let me emphasize that stimuli that can instantiate a phenomenal continuum
need not be pairwise indiscriminable. Since discriminable stimuli sometimes ap-
pear the same (e.g., in 25 percent of same/different trials), there is nothing to pre-
vent a series of discriminably different stimuli from instantiating a phenomenal
continuum for a given observer on a given occasion. (Also there is nothing to
prevent an indiscriminability series from failing to instantiate one.) Let me say
that again: there is nothing to prevent a series of discriminably different stimuli from
7
Martin clearly has the occurrent relation in mind in the second and third sentences of the
passage from 2004, 76 cited above; but I cannot find a consistent usage in the texts.
8
I borrow the expression “good straight look” from C. L. Hardin (1988).
Against Strong Content
184
x
x
FIGURE 7.1 Stimulus configuration for the first of two tasks in the experiment. Adjacent stimuli to
be judged (same/different) on a given trial are indicated by black dots. (For illustration in black
and white, I have marked the endpoints of the hue series with ‘x’.)
9
Delwin Lindsey, Department of Psychology, Ohio State University; Angela Brown, School
of Optometry, Ohio State University. I present these results also in Raffman 2011, in a much more
detailed discussion of (in)discriminability.
10
There were 21 physically distinct stimulus values. If we label the 21 values as a‒u, their order in
the circle can be specified as a, a, b, b, c, c, and so on. Consecutive trials then involved the pairs a/a, a/b,
b/b, b/c, c/c, etc. The “redundant” pairs (a/a, b/b, etc.) tested for false alarms, viz., “different” responses
to identical stimuli. The latter data are irrelevant to the present discussion.
Against Strong Content
186
FIGURE 7.2 Stimulus configuration for the second task in the experiment, performed on trials in
which adjacent patches had been judged same in the first task. (This included all but one trial by
one subject.) Subjects adjusted the hue of the central disk to match the hue of the two patches
indicated by black dots.
the circle, as in figure 7.2. The subject then adjusted the hue of the disk by moving
the computer mouse back and forth until the hue of the disk matched the hue of
the two patches. The disk then disappeared and the next trial began. (Both the
initial hue of the disk and the relationship between the hue of the disk and the
mouse position were randomized from trial to trial.) Subjects went around the
circle twice. At the end of the experiment we asked roughly half of the subjects if
they had noticed any changes in the colors of the patches during the experiment.
All said “no.”
What we found was that even though all of the patches were in view
throughout, and the members of every pair were judged “same” by every subject
on almost every trial, subjects’ settings of the disk progressed more or less system-
atically with the physical reflectances of the patches (figure 7.3). In other words,
subjects matched the pair #2/#3 to a longer wavelength than the pair #1/#2, the
pair #3/#4 to a longer wavelength than the pair #2/#3, and so on.11 This means that
patch #2 was matched to a different wavelength when it was compared to #1 than
when it was compared to #3; patch #3 was matched to a different wavelength when
compared to #2 than when compared to #4; and so forth—again, even though all
of the patches were in view. (The graphs contain more than 41 data points because
subjects went around the circle twice; hence, pairs that were judged “same” both
times received two disk settings.) The graphs show fairly steady progression of the
disk settings as subjects progressed through the pairs of patches, for both the phys-
ically identical and physically different pairs.12 Figure 7.4 shows the disk settings
averaged across all subjects.
11
For convenience I use the term “wavelength,” but strictly speaking it is incorrect. The stimuli were
mixtures of broadband lights, and neither the primaries nor the mixtures had a defined wavelength.
12
This result suggests that subjects may have been matching the hue of the disk to the mean
physical value of the two patches in each pair.
Disjunctivism, Discrimination, and Categorization 187
(a) 1000
950 Physically same
900 Physically different
Disk setting in arbitrary units
850
800
750
700
650
600
550
500
450
400
0 5 10 15 20 25 30 35 40
Stimulus pair
(b)
1000
950
Same
900 Different
Disk setting in arbitrary units
850
800
750
700
650
600
550
500
450
400
0 5 10 15 20 25 30 35 40
Stimulus pair
FIGURE 7.3 Disk settings (in arbitrary units) for two subjects, on stimulus pairs judged “same”.
No doubt these data permit multiple interpretations. But they provide at least
some support for the idea that individual patches changed their hue appearances in
their different pairwise comparisons; the patches looked different from one com-
parison to another, but so subtly that subjects could not notice it. Plausibly, each
pairwise comparison required a distinct attentional act, and the stimuli changed
the way they looked with each shift of attentional focus. The change is so slight—
perhaps something like an extremely fine-grained Gestalt shift in hue—that sub-
jects could not notice it, and so could not report it.13 Although the phenomenology
13
Raffman (2000) contains a more detailed discussion of the possibility of Gestalt hue shifts in a
sorites series.
Against Strong Content
188
1000
900
800
700
600
500
0 5 10 15 20 25 30 35 40
FIGURE 7.4 Disk settings averaged across all subjects.
is very difficult to describe, it seems unlikely that subjects were undergoing merely
a subliminal or unconscious sort of perception, since making a same/different
comparison and setting the hue of the disk required sustained conscious attention.
Borrowing some nice terminology from Fred Dretske (2004), perhaps we can say
that subjects saw different hues, consciously saw them, but could not see that they
were different; or that subjects saw different hues, but could not see the difference
in hue. Or perhaps they saw different hues but could not see the changes in hue.14
However the phenomenology should be described, our results appear to
support the hypothesis that unnoticeable changes occur in the appearances of
individual stimuli in a phenomenal continuum. These subtle changes, not any
nontransitivity, make phenomenal continua possible. If that’s right, then neigh-
boring stimuli in a phenomenal continuum do indeed appear the same in pairwise
comparisons, but the instability of their individual appearances across different
pairwise comparisons defeats the nontransitivity claim. Although any indiscrim-
inability series can instantiate a phenomenal continuum, such a series is only
physically, not phenomenally, stable. A claim of nontransitivity is true of the statis-
tical relation of indiscriminability because the relevant physical properties of the
stimuli (wavelengths, frequencies, etc.) in an indiscriminability series are stable.
14
Dretske introduces this terminology in order to characterize the experience of change
blindness. The present results may be reminiscent of change blindness, but subjects experiencing the
latter effect are able to notice the change when their attention is explicitly directed to it, whereas that
is unlikely to be the case in our study. Also, our experimental condition involved no visual disruption.
In general, subjects who experience change blindness fail to notice large changes in visual scenes when
the changes occur during a visual disruption such as a saccade or blink or a cut in a film (though see
Simons et al. 2000). For example, viewers in one experiment failed to notice that two people in a scene
had exchanged heads (Grimes 1996).
Disjunctivism, Discrimination, and Categorization 189
15
Of course, the experiment does not conclusively show this. Rather, it shows that we cannot
assume that subjects have a single, stable experience of B.
Against Strong Content
190
between some neighboring stimuli. They fail to consider the possibility suggested by
the experiment, viz., that such a progression can occur if individual stimuli subtly
shift their appearances, change hats, from one pairwise comparison to the next.
There need be no “discrimination,” explicit or implicit, between adjacent stimuli.16
Even if the worries about nontransitivity can be set aside, the relation of appearing
the same probably cannot do the philosophical work that Martin’s disjunctivist
requires of it. Let me try to say why.
Martin claims that “some [token] event is an experience of a street scene just
in case it couldn’t be told apart through introspection from a veridical percep-
tion of the street as the street.” What exactly is this supposed to mean? The locu-
tion “couldn’t be told apart through introspection from a veridical perception of
the street as the street” must be short for something like “couldn’t be told apart
through introspection from a veridical perception of the street as the street, were
the event in question to occur simultaneously with such a perception in a same/
different comparison.” Consider then an event e and some veridical perception v
of the street as the street. The trouble with Martin’s view is that, if the two events
do not actually occur simultaneously (or in immediate succession), the imagined
comparison would require a temporal relocation of at least one of the two. On the
plausible assumption that the time of occurrence of an experience is partly con-
stitutive of its being that very experience, the imagined comparison could not be
a comparison of e and v.17 Perhaps Martin imagines that two token experiences
couldn’t be told apart just in case, if two other token experiences of the same types
as e and v, call them e* and v*, were to occur simultaneously or in immediate
succession in a pairwise comparison, the subject would judge e* and v* to be the
same. But this strategy employs the very notion—that of experiences being of the
same type—that he is supposed to be explicating. Furthermore, the co-occurrence
of e* and v* would, or at least could, constitute a context relevantly different from
the contexts in which e and v occur, thereby threatening the type-identity of e* and
v* to e and v.
The upshot, I think, is that we cannot meaningfully talk about whether two
experiences could be told apart were they to occur simultaneously or in imme-
diate succession. Martin’s and Siegel’s talk of the (in)discriminability of experi-
ences is, in the loose and popular sense, a category mistake. Only physical stimuli,
whose identity conditions do not essentially involve states (e.g., attentional foci) of
16
Sturgeon (2006) and Conduct (2010), among others, offer alternative responses to the alleged
nontransitivity problem. Insofar as the approach I take is grounded in the perceptual psychology of the
situation, it may be better motivated than these others.
17
Siegel is aware of this problem (2004, p. 109).
Disjunctivism, Discrimination, and Categorization 191
18
See above. Lest there be any confusion: the distinction between discriminatory and categorical
judgments is wholly orthogonal to the distinction between statistical and “occurrent” (in)discriminability
relations (i.e., between [in]discriminability proper and appearing the same.)
Disjunctivism, Discrimination, and Categorization 193
given the reference to a determinate shade? Given the limited nature of perceptual
memory, we cannot recognize or type-identify instances of (e.g.) red28 as such; but
can we nevertheless say that an event is an experience as of red28 just in case intro-
spectively it seems category-identical to a veridical perception of red28?
Consider that we already have in hand non-discriminatory identity condi-
tions for, and can recognize upon inspection, the four so-called unique hues.
These latter are pure instances of the four chromatic categories—red containing
no blue or yellow, green containing no blue or yellow, blue containing no red or
green, and yellow containing no red or green. In addition, a perfectly balanced
orange can be defined as the hue (determinate shade) containing equal amounts
of red and yellow, a perfectly balanced cyan as containing equal amounts of blue
and green, and so forth. Using a technique called magnitude estimation, subjects
can reliably identify the unique hues as the hues containing 0 percent of any other
chromatic category, the balanced binaries as containing 50 percent of each compo-
nent category, and so on (e.g., Ebenhoh and Hemminger 1981; Gordon et al. 1994).
Of course, the extent of our ability to identify or categorize determinate shades
in this way cannot be unlimited: for example, presumably we cannot recognize as
such a shade that is 43 percent blue and 57 percent green as opposed to 44 percent
blue and 56 percent green—much less 43.5 percent blue and 56.5 percent green,
etc. Presumably this is only for lack of the requisite resolving power, though.
Unless we have independent reason to think that our experiences of non-unique,
non-balanced determinate hues are relevantly different from our experiences of
the unique hues and balanced binaries, we can plausibly treat our experiences of
the non-unique hues as being of a kind with our experiences of the unique and
balanced ones, and regard them as similarly type-identifiable at least in principle.19
The idea then would be to say that an event is an experience as of red28 just in
case introspectively it seems category-identical to a veridical perception of a shade
that is (e.g.) 55 percent red and 45 percent yellow.20 Whether such an approach
would satisfy the disjunctivist, I am not certain. Even granting that we can iden-
tify determinate shades in the way just outlined, I am not sure whether identity
conditions expressed in terms of percentages of different chromatic components
can, as Siegel puts it, provide “a way to characterize the . . . pairs of perceptions
and hallucinations . . . [that] raise the question whether they share a fundamental
property” (2004, p. 108). While the proposed strategy might alleviate some of the
problems associated with identity conditions expressed in terms of same/different
judgments, its reliance on the subject’s ability to recognize determinate shades
may veer too close, for the disjunctivist’s liking, to injecting intentional contents
19
It must be remembered that, even if we can individuate determinate shades in the way I am
proposing, these qualities are highly unstable properties of objects; in particular, objects may change
their “hats” from one pairwise comparison of stimuli to another.
20
As before, this category identity would be something the subject can descry in judging a single
event; no hypothetical comparison of simultaneous events would be required.
Against Strong Content
194
into the definition of an experience.21 From the disjunctivist’s point of view, replac-
ing discrimination with categorization in the way I have suggested may be simply
replacing one set of problems with another.22
References
21
Another potential problem is that no analogous account could be given for pitch perception
(among other things). Listeners lacking perfect pitch, i.e., most listeners, have no mnemonic anchors
for pitch that are analogous to our mnemonic anchors for the unique hues. Perfect pitch just is the
ability to hear (remember) pitches the way any normally sighted person sees (remembers) colors.
22
I am grateful to Mohan Matthen and Susanna Siegel for extremely helpful comments.
Disjunctivism, Discrimination, and Categorization 195
Raffman, D. (2011). Vagueness and cognitive science. In G. Ronzitti (Ed.), Vagueness: A Guide
(pp. 107–122). Dordrecht: Springer.
Siegel, S. (2004). Indiscriminability and the phenomenal. Philosophical Studies, 120(1–3),
91–112.
Simons, D., Franconeri, S., & Reimer, R. (2000). Change blindness in the absence of a visual
disruption. Perception, 29, 1143–1154.
Soteriou, M. (2009). The disjunctive theory of perception. In Stanford Encyclopedia of
Philosophy. Metaphysics Research Lab, CSLI, Stanford University.
Sturgeon, S. (2008). Disjunctivism about visual experience. In A. Haddock and F.
MacPherson (Eds.), Disjunctivism: Perception, Action, Knowledge. Oxford: Oxford
University Press.
Williamson, T. (1990). Identity and Discrimination. Oxford: Basil Blackwell.
PART THREE
Reconciliatory Views
8
I am grateful to Bill Brewer, Todd Ganson, and an anonymous referee for Oxford University
Press for detailed written comments. An early version of this paper was presented at CUNY and the
University of Miami. Thanks also for many helpful discussions at those occasions.
1
For austere representationalist views, see McGinn (1982), Davies (1992), Tye (1995), Lycan
(1996), and Byrne (2001) among many others. For austere relationalist views, see Campbell (2002),
Travis (2004), Johnston (2004, 2006), Brewer (2006), Fish (2009), Genone (forthcoming), and Raleigh
(forthcoming) among others. Martin (2002a, 2004) argues against any view on which experience can be
analyzed in terms of a propositional attitude and a content, leaving open the possibility that experience
could have content without the subject standing in a propositional attitude to that content. Since he
does not outright deny that experience has content, I will discuss his view only to the extent that his
positive view of perceptual experience is structurally similar to that of austere relationalists. Campbell
(2002) calls his view the “relational view,” Martin (2002a, 2004) calls his “naïve realism,” while Brewer
(2006) calls his the “object view.” I will refer to the view with the label “austere relationalism” since
the most distinctive features of the view are arguably the central role of relations between perceiving
subjects and the world as well as its austerity: the view is austere insofar as it denies that experience has
any representational component. There is room in logical space to reject representationalism without
endorsing a relationalist view. For such a view, see Gupta (2012). 199
Reconciliatory Views
200
Let the Content Thesis be the thesis that perceptual experience is fundamentally a
matter of representing the environment as being a certain way.
Content Thesis: Perceptual experience is fundamentally a matter of represent-
ing the environment as being a certain way.
As austere relationalists point out, this thesis is rarely argued for. While many
views have been defended that endorse it, more often than not such views simply
assume that experience is representational and proceed to argue for one particu-
lar way of understanding its content. Following Campbell (2002), I use the label
“the representational view” or “representationalism” for any view that endorses the
Content Thesis.2
There are three critical choice points for any representationalist view of per-
ceptual experience. One choice point is how to understand the relationship be-
tween the experiencing subject and the content of her experience. In order to
2
It is important to distinguish this view from the more specific view according to which the
sensory character of experience supervenes on or is identified with its content. Such views are
sometimes labeled “representationalism” rather than the more traditional “intentionalism.” I will
reserve “representationalism” for any view that endorses the Content Thesis. “Representationalism,” so
understood, is neutral on the relationship between content and sensory character.
The Relational and Representational Character of Perceptual Experience 201
3
As I have argued elsewhere (see my 2011), some accounts of perceptual content fall prey to the
austere relationalist objections; others (in particular the view I defend which acknowledges mutual
dependence of the relational and representational character of perceptual experience) arguably do not.
4
For a detailed discussion of the relation between the content of pictures and the content of
experiences and mental states more generally, see Crane (2009).
5
Byrne (2009) and Siegel (2010) have presented arguments in support of the view that experience
has content, but arguably their arguments do not establish more than the Association Thesis.
Reconciliatory Views
202
to the sense in which one might say that belief is a matter of standing in the
believing relation to the content of the belief. English does not have a word to
denote such a perceptual attitudinal relation. Byrne (2009, p. 437) calls the rela-
tion the ex-ing relation; Pautz (2010, p. 54) calls it the sensorily entertaining re-
lation; Siegel (2010, p. 22) calls it the A-relation. The Propositional Attitude Thesis
is a version of the Content Thesis. However, it is important to keep in mind that
we can accept the Content Thesis without accepting the Propositional Attitude
Thesis: The Content Thesis is committed neither to the content of experience
being a proposition nor to the experiencing subject standing in a propositional
attitude to the content of her experience.
An even more controversial version of the Content Thesis has it that the relation
between the experiencing subject and the content of her experience is an awareness
relation: The experiencing subject stands in an awareness relation to the content
or its constituents, such that this awareness relation grounds the sensory character
of the experience. This Awareness Thesis can be traced back to Russell (1913), who
argued that an experiencing subject stands in acquaintance relations to particulars
that in turn can be understood as the constituents of the proposition that character-
izes her experience.6 In the tradition of Russell, some representationalist views are
formulated in a way that suggests a commitment to the Awareness Thesis. While
the Awareness Thesis entails the Content Thesis, the converse is not the case: we
can accept the Content Thesis without accepting that perceivers stand in any kind
of awareness relation to the content of their experience. Experience can be under-
stood to have content in that the experiencing subject represents her environment
by employing perceptual capacities without the subject standing in an awareness
relation to the content yielded by employing those capacities.
The Awareness Thesis and the Propositional Attitude Thesis carry controversial
commitments that the Content Thesis does not entail. As I will show, we can accept
that perceptual experience is fundamentally a matter of representing one’s envi-
ronment without being committed to these more contentious theses. The Content
Thesis is neither committed to the thesis that the content of experience is a prop-
osition, nor is it committed to the thesis that the experiencing subject stands in
an awareness relation to the content of her experience. This is important since at
least some arguments against the Content Thesis assume that this thesis entails
those more controversial theses.7 Such arguments lose their grip, if one recognizes
that the Content Thesis does not carry the commitments of the more controversial
theses.
So far, we have distinguished different ways of understanding the relation
between the experiencing subject and the content of her experience. A second
6
One could argue however that on Russell’s view, acquaintance with particulars and universals
is more basic than any contentful mental state in that acquaintance with particulars and universals
explains how it is possible to entertain the relevant contents. Thanks to Bill Brewer for pressing me on
this point.
7
See, for example, Travis (2004).
The Relational and Representational Character of Perceptual Experience 203
choice point one faces is how to understand the relationship between the content
and the sensory character of perceptual experience. One might identify content
and sensory character. Alternatively, one might deny identity but maintain ei-
ther that content is grounded in sensory character or that sensory character is
grounded in content. Or, one might treat content and sensory character as inde-
pendent elements of experience, thereby denying that there is any identity or
grounding relation between content and sensory character.
A third choice point is how to understand the nature of perceptual content. This
choice point contains several levels. One is whether the content is understood in
terms of a Russellian proposition, a possible world proposition, a Fregean sense, or
in some other way. A second is whether perceptual content is conceptually or non-
conceptually structured.8 A third is whether or not the content of experience is prop-
ositionally structured. A fourth is whether the content is (at least in part) externally
individuated and so dependent on the experiencer’s environment, or alternatively
only internally individuated and so independent of the experiencer’s environment.
I will consider the final point in more detail, since the question of whether
perceptual content is environment-dependent is crucial in the debate on whether
experience is fundamentally representational, relational, or fundamentally both
relational and representational.9 According to austere representationalism, the
content of experience is internally individuated in the sense that it is indepen-
dent of the environment of the experiencing subject.10 The view is austere since
it leaves no significant room for a relational component. The only difference be-
tween subjectively indistinguishable experiences in distinct environments is a
difference in the causal relation between the experiencing subject and her envi-
ronment. According to austere representationalists, this difference in causal re-
lation has repercussions neither for the content of the experiences nor for their
sensory character. On such a view, the content of experience can be analyzed in
terms of existentially quantified content of the form that there is an object x that
instantiates a certain property F: (∃x)Fx. So experience represents only that there
is an object with the relevant properties in the external world. No element of the
content depends on whether there is in fact such an object present. Austere repre-
sentationalism has it that the content lays down a condition that something must
8
The debate on whether perceptual content is conceptually or nonconceptually structured is
sometimes understood as a debate about whether perceptual content is structured by Fregean concepts
and not just properties and objects. On this understanding, the first and second level distinguished
above collapses. However, there are ways to understand perceptual content as structured by modes of
presentation without committing oneself to the idea that the content is thereby conceptually structured
(see, for example, Schellenberg 2013). Therefore, I distinguish the two debates. Thanks to Todd Ganson
for pressing me on this point.
9
Nanay (forthcoming) argues that the debate between representationalists and relationalists is
best understood as a debate not about what is fundamental in an account of perceptual experience but
rather as a debate about the individuation of perceptual states.
10
McGinn (1982), Davies (1992), Tye (1995), Lycan (1996), and Byrne (2001) among others
have defended views that are committed to perceptual content being independent of the experiencer’s
environment.
Reconciliatory Views
204
Everything I will say about this case needs to be modified only slightly to fit with other
versions of austere relationalism. Given this case, the austere relationalist thesis can be
articulated in the following way: A subject perceives a particular white cup only if she
is perceptually related to that particular white cup; no appeal to content is necessary
to fully explain the nature of the subject’s perceptual experience. Being perceptually
related to a white cup may in turn be analyzed in terms of being perceptually related
to a cup instantiating whiteness, where the relevant object and property-instance are
collocated. More generally, subject S perceives object o as instantiating property F
only if S is perceptually related to o and an instance of F, where o and the instance of
F are collocated.
It will be helpful to make three clarifications about the view at stake. First, aus-
tere relationalists do not deny that beliefs and judgments are formed on the basis of
perception. So what is contentious is not whether perception brings about mental
states with content. The questions at stake are rather whether this content is an as-
pect of perception proper and whether the thesis that experience is representational
is fundamental in an account of perceptual experience. Second, austere relational-
ists do not contest that perception involves cognitive or neural processing that can
be characterized in terms of representations (Campbell 2002, 2010), but insist rather
that no appeal to content is necessary to explain the nature of the awareness of our
surroundings that we have as a consequence of this cognitive processing. So while for
example Campbell allows that representations play a role on a subpersonal level, he
denies that any appeal to representations is necessary to explain perception on a per-
sonal level. Finally, austere relationalists need not deny that we can articulate proposi-
tions to express what we experience. Acknowledging that a subject can articulate such
propositions entails no commitment to positing that her experience itself has the con-
tent articulated. It might just be that the propositions articulated are merely associ-
ated with the experience. Austere relationalists can accept the Association Thesis. So in
order to establish the Content Thesis, we cannot simply appeal to the fact that we can
articulate propositions to express what we experience. Appealing to such a fact would
merely establish the Association Thesis. We need to show that these propositions or
contents are a proper aspect of experience and indeed that they are a fundamental
aspect of experience.11
I will present my argument for the Content Thesis in two stages. I will first put
forward the Master Argument for perceptual content. The Master Argument is
11
For an argument that disposing of perceptual representations is inconsistent with empirical
findings about dorsal perception and about the multimodality of perception, see Nanay (this volume).
For a discussion of how the ventral and the dorsal stream work together in visual experience, see also
Wu (forthcoming). For a critical discussion of recent representationalist views on empirical grounds,
see Bronner, Kerr, and Ganson (forthcoming).
The Relational and Representational Character of Perceptual Experience 207
compatible with accepting only the Association Thesis, and so is not sufficient for
establishing the Content Thesis. However, we will need it to establish that thesis,
since it clarifies the relevant notion of content. In the next section, I will put for-
ward the Argument for Relational Content, which builds on the notion of content
established by the Master Argument. By arguing that perception is fundamentally
a matter of representing one’s environment, I will be in a position to conclude the
Content Thesis. By arguing that perceptual content is best understood to be inher-
ently relational, I will be in a position to conclude that perceptual experience is
fundamentally both relational and representational.
The mere fact that the environment sensorily seems a certain way when one per-
ceives supports a standard notion of perceptual content. The Master Argument for
perceptual content goes as follows:
12
While in my (2011) I formulate this argument in terms of “aware” and “seems,” I here speak
of “sensorily aware” and “sensorily seems.” This clarifies the relevant notion of “aware” and “seems.”
Furthermore, while in my (2011) I formulate this argument by saying that the subject is perceptually
related to the world, I here formulate the same argument by saying that the subject is perceptually
related to her environment. This clarifies which part of the world the subject is perceptually related to.
Brewer’s response to my argument is in no way affected by these two clarifications.
Reconciliatory Views
208
Brewer accepts the first two premises of the argument. It is not contentious that
if we are perceptually related to our environment, then we will be sensorily aware
of that environment (P1). Moreover, it is not contentious that if we are sensorily
aware of our environment, then our environment will sensorily seem a certain way
to us (P2).
We can moreover recognize P3 to be true, if we recognize that there is a
notion of content on which the content of experience corresponds to the way the
environment sensorily seems to the experiencing subject. Let’s call this connec-
tion between content and the way the environment seems the seems-content link.
Since we are talking only of sensory seemings the relevant cases are constrained to
those in which a subject is hearing, seeing, smelling, touching, or experiencing the
environment in some other sensory mode, or a combination of sensory modes. If
we recognize the seems-content link, then the idea that the environment can seem
a certain way to a subject without her being in a contentful mental state becomes
impossible. But the idea that the environment can seem a certain way to a subject
without her being in a contentful mental state is precisely the idea that austere
relationalism relies on.
Recognizing the seems-content link is compatible with accepting that any given
scene can be perceived in many different ways. The way the environment seems
to the perceiver may change from moment to moment even as her gaze remains
steady. Say she is looking at a pig. She can direct her attention at its shape, its color,
the texture of its skin, or any combination of these features. As her attention shifts,
her sensory character will change. One or more propositions can be associated with
every one of these phenomenal states and thus with every one of these ways that the
environment may seem to her. All of these propositions or sets of propositions are
equally legitimate contents of possible experiences she may enjoy while beholding
the pig. Nevertheless, at any given moment the environment will seem to her to be
one single way. This is all that we need to establish the seems-content link.13
Once one has recognized the seems-content link, only minor further com-
mitments are necessary to establish that the way the environment seems to an
experiencing subject is assessable for accuracy. In virtue of a subject perceiving
the environment, it seems a certain way to her. The way the environment seems to
a subject determines the way the environment would have to be for the content of
her experience to be accurate. The environment is either the way it seems to her
or it is different from the way it seems to her (P4). If the environment is the way it
seems to her, then the content of the experience is accurate. In all other cases, the
content of the very same experience is inaccurate. So if a subject has an experience
13
For a detailed discussion of how these premises must be interpreted depending on whether
one understands the seemings in question comparatively or noncomparatively, see my (2011). For
present purposes we can ignore this detail, since the Master Argument can be accepted regardless of
whether seemings are understood comparatively or noncomparatively. For the distinction between
the comparative and noncomparative use of appearance words, see Chisholm (1957, pp. 50–53) and
Jackson (1977, pp. 30ff.).
The Relational and Representational Character of Perceptual Experience 209
with a particular content, then this content is either accurate or inaccurate (P5). It
follows from this, together with P1 and P2 of the Master Argument, that if a subject
is perceptually related to the environment, then the way the environment seems to
her is assessable for accuracy. Together with P3, it follows that if a subject is per-
ceptually related to the environment, then the content of her experience is either
accurate or inaccurate.14
The Master Argument does not on its own establish the Content Thesis. After all, it
makes no claims about whether experience is fundamentally a matter of represent-
ing. On a weak reading of the Master Argument, it establishes only the Association
Thesis. In order to establish the Content Thesis, we need an additional argument.
One such additional argument is the Argument for Relational Content:
14
It will be helpful to make two clarifications about the thesis that the way the environment seems
to one determines accuracy conditions. First, there can be phenomenal differences between experiences
that are not a matter of how the environment seems to one, but rather a matter of how one experiences.
If I am shortsighted, my experience may be blurry, but I need not perceive the environment as being
blurry. I have argued that perceptual content corresponds to the way the environment seems to the
perceiver. This seems-content link is neutral on how those aspects of sensory character are accounted
for that do not pertain to the way the environment seems to the perceiver. Second, the environment is
arguably rarely and perhaps never the way it seems to us to be. We perceive plates to be round, although
their shapes are much more complicated. We see surfaces to be colored, but it has been argued that
surfaces do not have color properties. We see our environment to be populated by objects, but it has
been argued that there really are no objects or at least not the kind of objects that we seem to see. In
order to accommodate these phenomena, we need to loosen the notion of accuracy conditions in play
or alternatively we need to accept widespread but explicable perceptual error. For a detailed discussion
of this set of issues, see Pautz (2009) and Siegel (2010). If my argument for the thesis that experience
has accuracy conditions holds, then it holds regardless of what stance one takes on this set of issues.
Reconciliatory Views
210
The basic idea in support of this argument is that when we perceive, we employ
perceptual capacities by means of which we differentiate and single out particulars
in our environment. The relevant particulars are external and mind-independent
objects, events, property-instances, and instances of relations. Sensory seemings
are understood as individuated by employing such perceptual capacities in a sen-
sory mode, that is, modes such as seeing, hearing, touching, smelling, or tasting.
Employing such perceptual capacities in turn constitutes a mental state with
content.
Say we perceive a white cup on a desk. We employ our perceptual capacity to
discriminate white from other colors and to single out white in our environment.
Similarly we employ our capacity to differentiate and single out cup-shapes from,
say, computer-shapes and lamp-shapes. It is not clear what it would be to single
out an object in our environment without employing capacities of this kind. If this
is right, then it is in virtue of employing such capacities that we are in a sensory
state that is of a white cup.
Now how should we understand the capacities in play? They can be under-
stood as conceptual capacities or as nonconceptual capacities. There are pow-
erful reasons to understand perceptual content as nonconceptually structured.15
Therefore, I will focus on nonconceptual perceptual capacities. Indeed, I will focus
on the cognitively most low-level nonconceptual perceptual capacities, namely,
discriminatory, selective capacities. A discriminatory, selective capacity functions
to differentiate and single out, where singling out a particular is a proto-conceptual
analog of referring to a particular.16 So if we possess the discriminatory, selective
capacity that functions to differentiate and single out red, we are in a position to
differentiate instances of red from other colors in our environment and to single
out instances of red. More generally, to possess a discriminatory, selective capacity
is to be in a position to differentiate and single out the type of particulars that the
capacity concerns, were one related to such a particular. So if we possess such a
15
For discussion of nonconceptual content, see Peacocke (1998), Heck (2000), and Speaks (2005).
For a recent defense of the idea that perceptual content is conceptually structured, see Glüer (2009)
and Bengson et al. (2011).
16
In some cases, a discriminatory capacity may also function to type the kind of particulars that
the capacity concerns, but this is not an essential feature of the capacities in play.
The Relational and Representational Character of Perceptual Experience 211
17
For an alternative way of avoiding the pitfalls of both austere representationalism and austere
relationalism, see Dorsch (2013).
18
For a detailed development of the semantic nature of such token contents, see my (2010). For a
helpful discussion of the problems of factorizing mental content into internal and external components,
see Williamson (2000, 2006). See also Burge (2010).
The Relational and Representational Character of Perceptual Experience 213
19
For discussion, see Smith (2007), Batty (2010, 2011), and Fulkerson (2011). For discussions
of whether auditory experiences have attributional structure, see Nudds (2001), O’Callaghan (2010),
Ivanov (2011), Phillips (2013), and Matthen (forthcoming). See Macpherson (2011) for different ways
of individuating the senses.
Reconciliatory Views
214
In his book Perception and its Objects, Brewer responds to my argument. His
focus is on the first part of my Master Argument. He accepts the first two prem-
ises, but rejects the third premise of the argument. Recall that P3 has it that if a
subject’s environment seems a certain way to her, then she has an experience with
content C, where C corresponds to the way her environment seems to her.
Brewer notes that the Master Argument can be understood in ways compatible
with his view. As I have been careful to note, the Master Argument itself is com-
patible with just the Association Thesis. Accordingly, Brewer considers a strength-
ening of the Master Argument on which its conclusion is in genuine tension with
his view. He does so by replacing my P3 with a stronger thesis, that I will call P3*.
P3*: “the idea of a person having an experience whose most fundamental
nature is to be elucidated in terms of some kind of representational content
C” (2011, p. 61).
My P6‒P9 establish P3*, so my Argument for Relational Content is built into P3*.
Brewer rejects P3* on the following grounds:
I simply deny that it follows from the fact that there are truths of the form ‘o
looks F’ that apply to a person S in virtue of her perceptual relation with o,
that the most fundamental nature of that experience with the representational
content (of some kind) that o is F. On my interpretation of Schellenberg’s
Master Argument, this is the transition articulated explicitly by [P3*]. The
account of looks offered in ch. 5 proves that [P3*] is false on this interpre-
tation. For I explain there precisely how various looks claims apply to S in
virtue of her perceptual relation with the world around her without assuming
that the very nature of that perceptual relation is itself to be characterized in
terms of any corresponding worldly representational content. The perceptual
relation between perceivers and the mind-independent physical objects in
the world around them is on that account more basic than any such repre-
sentational contents and grounds the truth of the looks claims that perfectly
reasonably inspire talk of perceptual representation (2011, p. 60).
In a nutshell, his objection to my argument is that the nature of the perceptual
relation should not be understood as fundamentally involving representational
content. On Brewer’s view, the nature of perceptual experience is fundamentally
only relational.20 Brewer does not consider the prospect that perceptual experience
could be fundamentally both relational and representational. He seems to assume
that these options are exclusive. One of my main points is that these options are
20
Brewer focuses exclusively on the objects to which the perceiver is perceptually related, while
I focus on the environment to which the perceiver is perceptually related. The environment contains
objects, property-instances, relations, and events. This is a significant difference between Brewer’s
approach and my own. It can however be ignored for present purposes.
The Relational and Representational Character of Perceptual Experience 215
21
I would also quibble with Brewer’s wording when he characterizes representationalism as
endorsing the thesis that “the very nature of that perceptual relation is itself to be characterized in
terms of any corresponding worldly representational content” (2011, p. 60). I would not say that the
perceptual relation can be understood in terms of representational content, but only that perceptual
experience—rather than the perceptual relation—can be understood (at least in part) in terms of
representational content.
22
For a discussion of the relation between mental content and linguistic meaning, see Speaks
(2006). On perceptual reports, see Brogaard (forthcoming).
Reconciliatory Views
216
seeing o itself does not consist in the accuracy of those mental contents or can be
fruitfully illuminated by listing the facts that o looks F1, o looks F2, . . . ., o looks Fi,
etc., or the fact that it visually seems to S that o is F1, o is F2, . . . , o is Fi, etc. S’s see-
ing o, her perceptual experiential relation with that particular mind-independent
physical object is more basic than any such facts and is what grounds the accuracy
of all those mental contents.
In response to this rephrased version of Brewer’s argument, we can say that we
can accept that S seeing o is more basic than the accuracy of the mental content.
However, we can accept this without accepting that S seeing o is more basic than the
fact that S has an experience, which is fundamentally a matter of representing o. So
we can accept that S seeing o is more basic than the accuracy of the mental content,
while acknowledging that S seeing o is not more basic than S representing o. I argue
that S being perceptually related to o and S representing o are equally fundamental.
A further central reason for Brewer to reject representationalism is that
according to him representationalism cannot account for illusions. Brewer focuses
on the Müller-Lyer illusion, an illusion prompted by two lines that in fact have
the same length, but that seem to have different lengths due to outward looking
hashes on one line and inward looking hashes on the other line. He discusses sev-
eral ways a representationalist could analyze what the content of such an illusion
could be (2011, pp. 65–69). For the sake of argument, I will assume that the ways
of accounting for illusions in a representationalist framework that Brewer con-
siders do not work for the very reasons that Brewer cites. Rather than take issue
with Brewer’s arguments, I will put forward a way that a representationalist could
account for illusions that Brewer does not consider.
First, consider perceiving a cup at an angle. One way of analyzing what we rep-
resent when we perceive a cup at an angle is that we represent the shape of the cup
in two ways: <the cup has shapeʹ, the cup has shapeʹʹ>. The single primed property
is an intrinsic property and the double primed property is a situation-dependent
property, that is, it is a property that is determined by one’s location and the intrinsic
shape of the cup. More specifically, a situation-dependent property is a (noncon-
stant) function of an intrinsic property and one or more situational features, that
is, features of the environment that determine how objects are presented such as
the lighting conditions and the subject’s location in relation to perceived objects.
This means that fixing the intrinsic properties of an object and the situational fea-
tures fixes the situation-dependent properties. Furthermore, situation-dependent
properties are ontologically dependent on and exclusively sensitive to intrinsic
properties and situational features. I understand the intrinsic properties, of say a
white cup, to include among other properties, the shape and size of the cup. More
generally, intrinsic properties of an object are the properties that an object has re-
gardless of the situational features. They are the properties that an object has that
do not depend on the object’s relations to other individuals distinct from itself. 23
23
For a development of the notion of situation-dependent properties, see my (2008). For a critical
discussion, see Jagnow (2012) and Madary (2012).
The Relational and Representational Character of Perceptual Experience 217
While the single and double primed properties are different kinds of proper-
ties, a perceiving subject need not represent these metaphysical facts. The perceiv-
ing subject may just represent that the cup has a certain shape in one respect and a
different shape in another respect. So perception need not represent the metaphys-
ical basis of the distinction between shapeʹ and shapeʹʹ. For perceptual content to
be consistent, it is sufficient that a distinction is represented. It is not necessary
that it is represented what the metaphysical basis for that distinction is.
Now consider the Müller-Lyer illusion. One could say that the two lines that
prompt the Müller-Lyer illusion have the same length in one respect and have
different lengths in another respect. So one can distinguish two conceptions of
length: Call these lengthsʹ and lengthsʹʹ. In light of this distinction, we can say that
when we are perceptually related to the two lines, we represent <A and B have
the same lengths′, A and B have different lengthsʹ′>. The primes mark that one
distinguishes between different respects. Due to the primes, the content of our
experience is not inconsistent. In the case of the Müller-Lyer illusion, the sense
in which the lines look different in length cannot in any straightforward way be
analyzed in terms of situation-dependent properties. After all, there is no situ-
ational feature in the environment in virtue of which the lines look different in
length. However, the way in which we represent the difference between intrinsic
and situation-dependent properties can be exploited for an analysis of how we
represent the lines. As in the case of seeing the cup, we can say that the single and
double primed properties lengthsʹ and lengthsʹʹ are different kinds of properties,
but that a perceiving subject need not represent these metaphysical facts. For per-
ceptual content to be consistent, it is sufficient that a distinction is represented.
If this is right, then we have no reason to reject that experience is fundamentally
a matter of representing one’s environment to account for illusions. So again, we
have no reason to reject the Content Thesis.
References
Pautz, A. (2009). What are the contents of experiences? Philosophical Quarterly, 59, 483–507.
Pautz, A. (2010). Why explain visual experience in terms of content? In B. Nanay (Ed.),
Perceiving the World (pp. 254–310). Oxford: Oxford University Press.
Peacocke, C. (1998). Nonconceptual content defended. Philosophy and Phenomenological
Research, 63, 381–388.
Phillips, I. (2013). Hearing and hallucinating silence. In F. Macpherson & D. Platchias
(Eds.), Hallucination (pp. 333–360). Cambridge, MA: MIT Press.
Raleigh, T. (forthcoming). Phenomenology without representation. European Journal of
Philosophy.
Russell, B. (1913). Theory of Knowledge. London: Routledge.
Schellenberg, S. (2008). The situation-dependency of perception. Journal of Philosophy, 105,
55–84.
Schellenberg, S. (2010). The particularity and phenomenology of perceptual experience.
Philosophical Studies, 149, 19–48.
Schellenberg, S. (2011). Perceptual content defended. Noûs, 45, 714–750.
Schellenberg, S. (2013). Experience and Evidence. Mind, 487, 699–747.
Siegel, S. (2010). Do experiences have content? In B. Nanay (Ed.), Perceiving the World (pp.
333–368). Oxford: Oxford University Press.
Smith, B. (2007). The objectivity of taste and tasting. In B. Smith (Ed.), Questions of
Taste: The Philosophy of Wine (pp. 41–77). Oxford: Oxford University Press.
Speaks, J. (2005). Is there a problem about nonconceptual content? Philosophical Review,
114, 359–398.
Speaks, J. (2006). Is mental content prior to linguistic meaning? Noûs, 40, 428–467.
Travis, C. (2004). Silence of the senses. Mind, 113, 57–94.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press.
Williamson, T. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.
Williamson, T. (2006). Can cognition be factorized into internal and external components?
In R. Stainton (Ed.), Contemporary Debates in Cognitive Science (pp. 291–306). Oxford:
Blackwell.
Wu, W. (forthcoming). Against division: Consciousness, information, and visual stream.
Mind and Language.
9
Thanks to Adam Pautz, as well as audiences at York, Edinburgh, and Nottingham for helpful
220 questions and comments.
Experiential Content and Naïve Realism 221
hitherto offered. Second, pace Siegel, I think that Naïve Realism is compatible with
much stronger interpretations of the claim that experience has content.
In the first section of this chapter, after briefly arguing for the assumption that
experiential content is propositional, I’ll distinguish three interpretations of the
claim that experience has content (the Mild, Medium, and Spicy Content Views). In
the second section, I’ll flesh out Naïve Realism in greater detail, and I’ll reconstruct
what I take to be the main argument for its incompatibility with the Content Views.
The third section will be devoted to evaluation of existing arguments for the Mild
Content View (the arguments from accuracy and appearing), and the development
of what I take to be a stronger argument (the argument from belief generation). In
the final section, I’ll identify a flaw in the argument for the incompatibility of Naïve
Realism and the Content Views, which opens the door to a reconciliation.
(e.g., by claiming that experiential contents are typically much more specific or
finely-grained than the contents of beliefs).1 Another possibility is that the advocates
of non-propositional perceptual content have a particularly demanding conception of
what a proposition is in mind (as suggested in Byrne 2001, p. 201)—e.g., that they are
composed of Fregean senses. However, it is important to distinguish two questions:
(1) Do experiences have propositional content?
(2) If experiences have propositional content, what is the metaphysical struc-
ture of those propositions (e.g., Fregean, Russellian . . . )?
We must be careful not to simply assume an answer to (2) that supports a negative
answer to (1). One can show that experiences don’t have propositional content on
the basis of a particular answer to (2) only if one has an argument for giving that
answer rather than another—otherwise, it’s epistemically possible that experiences
have propositional content given a different theory of the metaphysical structure
of the relevant propositions.
Yet another potential source of resistance to identifying the way the subject of an
experience represents her environment as being with a proposition is the idea that an
experiencing subject represents things in the world, not propositions (cf. Crane 2009,
pp. 464–465). For example, one might insist that in having an experience of a yellow
crescent-shaped banana on my desk, I’m perceptually representing a banana (as well
as its color, shape, and location), not a proposition. However, this is a false contrast.
It’s not clear that there’s any daylight between, say, representing something as being
yellow crescent-shaped, and to my left, on the one hand, and representing the prop-
osition that something is yellow crescent-shaped, and to my left, on the other. Given
that a proposition is basically a way for the world to be, representing things in the
world as being certain ways is tantamount to representing a proposition.
I don’t take this brief case for skepticism about non-propositional experiential
content to be decisive, but unfortunately I don’t have the space to elaborate and
defend it. In any case, if you think there is such a thing as non-propositional expe-
riential content, then you should think of my aim as being to establish that Naïve
Realism is compatible with propositional experiential content.
Supposing that the content of experience is a proposition, which proposition
is the content of a given experience—say, my experience as of a yellow crescent-
shaped banana? As I see it, to say that an experience has content is to say at least
the following:
for any perceptual experience E,
(i) there is a proposition associated with E, and
(ii) this proposition captures the way things perceptually appear to the subject
in virtue of having E.
1
Thanks to Matt Nudds for pressing me to mention this potential motivation for denying
experiential content.
Experiential Content and Naïve Realism 223
For example, the content of my experience is (something along the lines of) the
proposition that there is a yellow crescent-shaped thing before me. These two
claims, (i) and (ii), are the bare minimum that one is committed to when one
says that experience has content. Let us call these two claims the Mild Content
View.2
Note that the Mild Content View is silent on whether experience is a prop-
ositional attitude—one could hold that there is a proposition associated with an
experience, but deny that the experience consists in the subject perceptually rep-
resenting it. On this sort of position, although the proposition figures in the char-
acterization of the experience from the theorist’s point of view, the subject of the
experience doesn’t bear any distinctively experiential psychological relation to it.
Alternatively, one could endorse the following claim:
(iii) perceptual experience consists in the subject perceptually representing
her environment as being a certain way.
Let us call claims (i)–(iii) the Medium Content View.
Note that the Medium Content View is silent on whether perceptual ex-
perience fundamentally consists in the subject perceptually representing her
environment as being a certain way—one could hold that experience is a prop-
ositional attitude, but not fundamentally so. To say that perceptual experience
fundamentally consists in personal-level psychological feature x is to say that
it has some or all of its other personal-level psychological features ultimately
in virtue of x. So, for example, one might hold that my experience of a yellow
crescent-shaped banana involves my perceptually representing my environment
as containing a yellow crescent-shaped thing, but that this fact isn’t the ultimate
personal-level psychological explanation of the why this experience naturally
generates the belief that there’s a yellow crescent-shaped thing before me, or
the phenomenal character associated with this experience, or any of the other
psychological features we’re trying to give an account of when we’re giving a
2
The Mild Content View adopts what Adam Pautz calls the appears-looks conception of
experiential content, and he argues that this conception trivializes the debate over whether experience
has content, as well as other related debates (2009, pp. 485–486). I won’t go into Pautz’s arguments here;
the crucial point is that (as he recognizes) one could try to avoid them by insisting on a distinction
between perceptual and epistemic appearances (more on this distinction in section 9.3). He seems to
think that this insistence must amount to holding that locutions of the form ‘it appears to S that p’
always pick out epistemic appearances, while perceptual appearances are only picked out by locutions
of the form ‘o appears F to S’. If this is right, then the appears-looks conception would still trivialize
the relevant debates (e.g., hallucinations would trivially lack perceptual content). However, one could
hold that locutions of the form ‘it appears to S that p’ are potentially ambiguous, sometimes picking
out epistemic appearances and sometimes picking out perceptual appearances. If that’s right, then the
debate is not trivial—as I will argue in section 9.3, it boils down to a debate over whether there is a
kind of perceptual appearance that ‘it appears to S that p’ is used to pick out. (Thanks to Adam Pautz
for pressing me to clarify this point.)
Reconciliatory Views
224
3
Here’s how the taxonomy of views I’ve just offered relates to some of the others in the
literature: Susanna Siegel (2010) distinguishes between the Content View and the Strong Content
View; the latter is more-or-less my Spicy Content View, and the former is essentially my Mild
Content View. (Siegel doesn’t single out what I’ve called ‘the Medium Content View’.) Pautz (2009)
distinguishes between appears-looks, accuracy, and identity conceptions of experiential contents;
the first is more-or-less my Mild Content View, and the last is basically my Spicy Content View. The
accuracy conception is an alternative—and in my view, inferior—way of formulating a Mild Content
View. (I won’t defend this claim here, although the reader might be able to discern my reservations
about the accuracy conception on the basis of the discussion of the argument from accuracy in section
9.3.) Finally, Susanna Schellenberg (2011, pp. 15–16) distinguishes between the association thesis and
the representation thesis. The former is basically my Mild Content View, while my Medium and Spicy
Content Views are different ways of spelling out Schellenberg’s representation thesis (both of which
are different from Schellenberg’s preferred way of spelling it out, which construes representation in
terms of employing concepts—my Medium and Spicy Content Views make no such commitment).
Schellenberg also identifies a view she calls the awareness thesis: basically, a view on which the content
of experience is a Russellian proposition, and the subject is aware of it in the sense that she literally
perceives its constituents (presumably, this is the only way to make sense of the idea that one could
literally perceive the propositional content of an experience). Since I will remain neutral in this chapter
on the nature of the propositions that are potential contents of experience, this view is beyond the
scope of this chapter.
4
See, e.g., Tye 2000 (Russellian content), Burge 1991 (Fregean content), Stalnaker 1984 (possible
worlds content), and Chalmers 2006 (a pluralistic view).
5
For negative answers to these questions, see (e.g.) Block 1990 and 1996; for affirmative answers,
see (e.g.) Tye 2000.
Experiential Content and Naïve Realism 225
take a stand on the presence of so-called “high-level” properties, such as natural kind
properties (e.g., being a banana)?6 I’m not going delve into any of these thorny issues
here, since doing so isn’t required in order to establish the main theses of this chapter;
namely, that experience has content in the weak sense, and that all three Content
Views are compatible with Naïve Realism.
6
For a case for the claim that natural kind properties (and other high-level properties) can figure
in the content of experience, see Siegel 2006; for objections, see (e.g.) Price 2009 and Logue 2013b.
7
This characterization of Naïve Realism is superficially different from some others found in the
literature (e.g., Fish 2009). See Logue (2013a) for discussion of how the characterization in the main
text captures the content of other typical formulations of the view.
8
One might suggest that the very notion of a veridical experience smuggles in an affirmative
answer to the question of whether experience has content, and so isn’t a notion I’m entitled to in this
dialectical context. The line of thought is this: A veridical experience is one in which things in the
subject’s environment appear to her to be a certain way, and they are that way (as the subject perceives
Reconciliatory Views
226
are experiences in which the subject perceives a thing o, and it appears to have a
property F even though the subject doesn’t perceive o’s F-ness. For example, consider
an illusion in which a green banana looks yellow as a result of unusual lighting condi-
tions. The subject of this illusion sees the banana, and it looks yellow to her, but not
in virtue of perceiving the banana’s yellowness—indeed, the banana isn’t yellow, so it
doesn’t instantiate any yellowness for her to perceive.
All Naïve Realists take their theory to apply to veridical experiences. However,
they divide when it comes to illusions. Some Naïve Realists restrict their theory
to veridical experience (e.g., Martin 2006). Others think that the theory can be
extended to illusions: one way to do this (very roughly) is to claim that illusion
fundamentally consists in the subject perceiving things and some of their proper-
ties, and that something’s illusorily appearing F consists in perceiving a property
distinct from F-ness that the thing does instantiate.9 The details of such an account
of illusion are beyond the scope of this chapter. For simplicity’s sake, I will restrict
my focus to Naïve Realism about veridical experience.
Why might a Naïve Realist be hostile to the claim that experience has con-
tent? Naïve Realism holds that veridical experience fundamentally consists in the
subject perceiving things in her environment; but on the face of it, this is perfectly
compatible with it having content as well. Just because Naïve Realists don’t typically
characterize veridical experiences as having experiential content, it doesn’t follow
that they couldn’t. Why would the Naïve Realist insist that veridical experience
consists in nothing more than the subject perceiving things in her environment?
I suspect that much of the Naïve Realist resistance to experiential content is rooted
in an argument of M. G. F Martin’s regarding a distinct but related issue—namely,
what a Naïve Realist should say about hallucinations. Let’s sketch this argument,
and explore its implications for the question of whether experience has content.
In his paper “The Limits of Self-Awareness” (2004), Martin argues that certain
accounts of hallucination don’t go well with Naïve Realism—in particular, what he
calls positive accounts. A positive account of hallucination is one that characterizes
it in terms that are independent of veridical experience. One example is the claim
that hallucination fundamentally consists in the subject perceptually representing
her environment as being a certain way (i.e., the Spicy Content View restricted to
hallucination). Note that this claim makes no reference at all to veridical experi-
ence. By contrast, a negative account of hallucination characterizes it in terms of a
the properties that appear to her to be instantiated by things in her environment). But, given what I’ve
said in the previous section, doesn’t the claim that things appear to her to be a certain way amount
to saying that her experience has content (the proposition that specifies how the things appear to her
to be)? In a word, no. As I’ll argue in the following section, the talk of appearance that figures in the
characterization of veridical experience can be understood in a distinctively perceptual sense or in an
epistemic sense. Only the former entails a commitment to experiential content, so one who denies
experiential content can make sense of the notion of veridical experience in terms of the latter.
9
For views roughly along these lines, see Brewer 2008, Fish 2009, Antony 2011, and Kalderon
2011.
Experiential Content and Naïve Realism 227
1. Suppose (for the sake of reductio) that veridical experiences have contents
(in the Mild sense that there are propositions that specify the way things
perceptually appear to the subject).
2. Naïve Realism is not explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (a
commitment of Naïve Realism).
3. The fact that a veridical experience has a given content is sufficient to
explain its epistemological, behavioral, and phenomenal features.
4. If veridical experiences have contents, Naïve Realism is explanatorily
redundant with respect to a veridical experience’s epistemological,
behavioral, and phenomenal features (from 3).
5. Naïve Realism is explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (from 1
and 4)
6. Contradiction (2 and 5); veridical experiences don’t have contents.
Of course, one could just as well re-frame this argument as a reductio of Naïve
Realism (by supposing premise 2 for the sake of reductio instead). But since our
issue is whether Naïve Realism is compatible with the Mild Content View, I will
grant premise 2 for the sake of argument. That leaves premises 3 and 4—what can
be said for them?
10
Strictly speaking, it’s not Naïve Realism that’s incompatible with a positive account of
hallucination—in principle, one could endorse Naïve Realism along with a positive account of
hallucination, and concede that Naïve Realism plays a very limited role in explaining the phenomena
that philosophical theories of experience are supposed to explain (e.g., the epistemological and
phenomenal aspects of experience). But it’s not clear why anyone would want to do that.
Reconciliatory Views
228
Premise 3 says that the fact that a veridical experience has a certain content
(e.g., that there is a yellow crescent-shaped thing before me) is sufficient to explain
the experience’s epistemological features (e.g., its tendency to generate the belief
that there’s a yellow crescent-shaped thing before me), its behavioral features (its
tendency to cause me to reach for the banana if I want to eat one), and its phenom-
enal features (i.e., what it’s like for me to see a yellow crescent-shaped thing). This
last plank of the premise is particularly controversial—it will be rejected by those
who hold that the phenomenal character of an experience doesn’t supervene on
its content. A popular reason for this rejection is the combination of externalism
about perceptual content with the alleged possibility of spectrum inversion. Given
externalism about perceptual content, one’s experience represents that something
is yellow in virtue of being a type of experience that is typically caused by yellow
things. Given the possibility of spectrum inversion, what it is like for one subject
to experience yellow things could be what it’s like for another subject to experi-
ence purple things. Putting the two together, two subjects could have experiences
with the content that there’s something yellow before them (because they’re hav-
ing experiences of the sort typically caused by yellow things), but what it’s like for
the subjects to have the experiences differs (because they are spectrally inverted
with respect to each other). If this is a genuine possibility, the phenomenal char-
acter of an experience doesn’t supervene on its representational content, and an
experience’s content isn’t sufficient to explain why it has the phenomenal character
it does—something more has to be said. For example, one might think we have
to appeal to intrinsically non-intentional qualia to fully capture experiential phe-
nomenology (Block 1996).
However, Naïve Realists are no friends of this kind of qualia, at least when it
comes to veridical experience. They typically hold that the phenomenal character
of a veridical experience is determined by the way the subject’s environment is—
that its phenomenal character is “constituted by the actual layout of the [environ-
ment] itself: which particular objects are there, their intrinsic properties, such as
color and shape, and how they are arranged in relation to one another and to [the
subject]” (Campbell 2002, p. 116). In short, standard Naïve Realism simply leaves
no room for phenomena like spectrum inversion that motivate the rejection of
the claim that phenomenology supervenes on content.11 Thus, most Naïve Realists
would be hostile to a rejection of premise 3 on the grounds just outlined. Again,
since our issue is whether Naïve Realism is compatible with the Mild Content
View, I will set aside this kind of reason for rejecting premise 3, and assume along
with most Naïve Realists that spectrum inversion isn’t a genuine possibility.
One might worry that no Naïve Realist would endorse premise 3. For such
an endorsement seems tantamount to admitting that there’s no motivation for
Naïve Realism—if experiential content really is sufficient to explain the episte-
mological, behavioral, and phenomenal features of experience, then we might as
11
For a version of Naïve Realism that does leave room for such phenomena, see Logue 2012b.
Experiential Content and Naïve Realism 229
well all embrace the Spicy Content View and call it a day.12 However, this worry
doesn’t withstand scrutiny. First, not all motivations for Naïve Realism claim that
it can explain a feature of veridical experience that the Spicy Content View cannot.
For example, Martin argues that Naïve Realism is required to account for sensory
imagination—he says that Naïve Realism and the Spicy Content View are on a par
with respect to accounting for perceptual experience (2002, p. 402). Premise 3 is
perfectly compatible with a motivation for Naïve Realism of this sort. Second, the
Naïve Realists who do think that there’s some feature of experience that only Naïve
Realism can explain don’t seem to have realized that a claim along the lines of
premise 3 is incompatible with their motivations for Naïve Realism. For example,
Bill Fish (2009) argues that only Naïve Realism can properly account for percep-
tual phenomenal character, but nevertheless wrongly takes his view to be subject
to Martin’s screening off argument. In short, some Naïve Realists can and do accept
premise 3, and at least some of those who shouldn’t don’t seem to have realized that
they shouldn’t. So the argument presented above is still a plausible reconstruction
of the Naïve Realist argument against experiential content.
So much for premise 3 (for the time being). What about premise 4? This
premise is a plausible consequence of premise 3. If the fact that a veridical experi-
ence has a given content is sufficient to explain its epistemological, behavioral, and
phenomenal features, then Naïve Realism drops out as redundant with respect to
accounting for veridical experience—if we can explain everything about the expe-
rience that a philosophical theory of perceptual experience is supposed to explain
just by saying that it has a certain content, then the Naïve Realist claim that it also
involves the obtaining of the perceptual relation between the subject and things in
her environment needlessly complicates our account of veridical experience. Of
course, it’s open to the Naïve Realist to insist that we should jettison the notion of
experiential content from our account instead.13 But the point is that it’s one or the
other—the marriage of experiential content and Naïve Realism appears to be an
unhappy one, since each steals the other’s explanatory thunder.
As I said before, I take it that something like the argument just presented
underlies much of the Naïve Realist hostility to experiential content.14 And on the
12
Thanks to Dave Ward for raising this important issue.
13
This is a disanalogy with Martin’s argument against positive disjunctivism. In that case, there is
a reason to prefer explanations in terms of experiential content: such explanations can provide a unified
account of the epistemological, behavioral, and phenomenal features shared by a veridical experience
and a hallucination. By contrast, a Naïve Realist explanation doesn’t apply in the case of hallucination.
Given that we should give a unified explanation of a phenomenon whenever possible, once we let
perceptual content into our account of perceptual experience, it affords the superior explanation of
the relevant phenomena. (In order to protect the explanatory power of Naïve Realism, Martin eschews
accounting for hallucination in terms of any positive experiential features like perceptual content—see
Martin 2004, pp. 71–72.)
14
This isn’t the only source of resistance—e.g., some have expressed doubts about whether we can
non-arbitrarily pin down the content of an experience (see Travis 2004 and Brewer 2006). Since my
main aim is to reconcile the Content Views with Naïve Realism, rather than to defend the former from
objections, I’ll set these doubts aside.
Reconciliatory Views
230
In this section, I will outline two arguments for the Mild Content View. I will argue
that, although they are sound, they aren’t likely to persuade Naïve Realists who are
convinced that their view is incompatible with the Mild Content View. So I will
offer a different argument for the Mild Content View that I take to be more dia-
lectically effective.
One argument for the Mild Content View is the argument from accuracy
(Siegel 2010, pp. 337–343; 2011, pp. 33–42). It runs as follows:
Siegel doesn’t rest her case for the Mild Content View on the argument from
accuracy, as she’s skeptical of premise 4. She notes that there’s no guarantee that an
experience’s accuracy conditions are conveyed to the subject, i.e., that the accuracy
conditions capture how things perceptually appear to her. As Siegel points out,
not all of the conditions under which an experience is accurate specify how things
perceptually appear to the subject. For example, take the trivial accuracy condition
of the experience’s being accurate—that one’s experience is accurate isn’t among
the ways things can perceptually appear to be (Siegel 2010, p. 344). Nevertheless,
given the claim that things perceptually appear to the subject of an experience to
Experiential Content and Naïve Realism 231
be a certain way, and that the experience has accuracy conditions, it’s natural to
identify the way things perceptually appear to the subject with at least some subset
of its accuracy conditions. That is, once we’ve recognized that experiences have
accuracy conditions, the burden is on the opponent of the Mild Content View to
explain why the way things perceptually appear to the subject isn’t identical to a
proposition specified by some subset of those conditions.15
Once the premises are tweaked so that the content of experience is specified
by a subset of an experience’s accuracy conditions, I believe the resulting argu-
ment is sound. But I don’t think it’s dialectically effective. Consider the situation
from the perspective of a Naïve Realist who’s convinced that her view is incom-
patible with the Mild Content View. The argument starts off with an appeal to
an alleged intuition—viz., that experiences are assessable for accuracy. Our Naïve
Realist thinks that endorsing this intuition is tantamount to giving up on her view,
so she’s well-advised to scrutinize this intuition carefully. And although I’m sym-
pathetic to the intuition, it’s not obvious that it’s worth hanging onto at any cost. If
I took myself to I have excellent reasons for endorsing Naïve Realism, and excel-
lent reasons to believe that Naïve Realism is incompatible with the Mild Content
View, I’d be willing to argue in the opposite direction from the falsity of the Mild
Content View to the falsity of the intuition. Philosophical reflection sometimes
recommends forsaking intuitions in favor of an error theory, and for our Naïve
Realist, this could be one of those cases. So, given the present state of the dialectic
(in which Naïve Realism is often taken by its proponents to be incompatible with
the Mild Content View), the argument from accuracy isn’t persuasive.
The route to the Mild Content View that Siegel does endorse is the argument
from appearing (Siegel 2010, pp. 345–354; see also Schellenberg 2011, pp. 718–720).
The argument goes roughly as follows:
15
I suspect Siegel would agree; it seems that her further argument from appearing (discussed
below) is intended to provide a way of specifying which subset of an experience’s accuracy conditions
yields its content.
16
Siegel and Schellenberg present much more detailed variants of this argument, but the details
wouldn’t affect the point I want to make here.
Reconciliatory Views
232
Mild sense). Prima facie, this argument is much more promising than the argu-
ment from accuracy. First, there’s no room to argue that the proposition specified
doesn’t capture the way things perceptually appear to the subject, since that’s what
specifies the proposition in the first place. Second, the starting point is much more
difficult to reject—one might be willing to reject the intuition that experiences
have accuracy conditions, but how can one deny that things perceptually appear
to be some way to one when one has an experience? Isn’t that a conceptual truth
if there ever was one?
Just as with the argument from accuracy, I think that the argument from
appearing is sound—it’s just not dialectically effective. Again, let’s think of the sit-
uation from the perspective of a Naïve Realist who takes her view to be incompat-
ible with the Mild Content View. If she’s right, the argument from appearing entails
the falsity of her view, so she’s well-advised to scrutinize its starting point. And (as
Siegel recognizes) there is some wiggle room here. The Naïve Realist might agree
that things appear to be some way to one when one has an experience, but only in
a sense of ‘appear’ that doesn’t entail the truth of premise 1.
It’s uncontroversial that things appear to the subject of an experience to be
some way in the sense that, in normal circumstances, the experience generates
the belief that things are that way. For example, when I have an experience of a
yellow crescent-shaped banana on my desk, there appears to me to be a yellow
crescent-shaped thing before me in the sense that I will form the belief that there
is a yellow crescent-shaped thing before me—at least as long as I’m rational, and
I don’t suspect that my experience has been generated in a non-standard way (such
as being the result of a hallucinogen). Such an appearance is an epistemic one. Let’s
say that it epistemically appears to a subject S that p just in case S is disposed to
believe that p solely on the basis of evidence (e.g., perceptual evidence), given that
S is rational and doesn’t suspect that the evidence is misleading. For example, it
epistemically appears to a subject of the Müller-Lyer illusion that the lines she’s
seeing are different lengths. Although a subject who’s in the know isn’t disposed
to believe that the lines are different lengths, she would be disposed to believe this
solely on the basis of her perceptual evidence if she didn’t know that her experi-
ence was misleading.
Now, the fact that things epistemically appear to the subject of an experience
to be some way doesn’t obviously entail that there is some other sense in which
things appear to the subject to be some way, associated with experiences instead
of beliefs—i.e., a distinctively perceptual appearance (see, e.g., Travis 2004; Brewer
2008 for claims along these lines). This is particularly clear on Naïve Realism, on
which veridical experience is fundamentally a relation to objects in one’s environ-
ment, unlike belief, which is fundamentally a relation to a proposition. According
to a Naïve Realist, the subject of a veridical experience perceives entities in her
environment (e.g., a yellow crescent-shaped banana), and this disposes her
to believe that her environment is a certain way (e.g., that there’s a yellow
crescent-shaped thing before her). And although there’s an epistemic appearance,
Experiential Content and Naïve Realism 233
in that the subject is disposed to believe on the basis of perceptual evidence that
there’s a yellow crescent-shaped thing before her, the Naïve Realist might insist
that that there’s no distinctively perceptual appearance in addition—that there is a
proposition that captures the way things perceptually, as opposed to epistemically,
appear to the subject. In short, the idea is that things appear to be some way to the
subject in virtue of having the experience, but the proposition that things are that
way is the content of perceptually-based belief, not experience proper—veridical
experience is a relation to concrete things in one’s environment, not abstract prop-
ositional contents.
Given that the claim that it perceptually appears to S that o is F is equivalent
to the claim that o perceptually appears F to S, to deny the former is to deny the
latter.17 So the Naïve Realist maneuver sketched in the previous paragraph would
be tantamount to denying that things perceptually appear to have properties—e.g.,
that the banana on my desk perceptually appears yellow. Furthermore, one might
think that a necessary condition of S perceiving o’s F-ness is that o perceptually
appears F to S—I wouldn’t be perceiving the yellowness of the banana on my desk
if it didn’t look (i.e., visually appear) yellow to me. If that’s right, the Naïve Realist
maneuver just outlined entails that the subject of a veridical experience simply
perceives things in her environment (e.g., bananas); strictly speaking, she doesn’t
perceive any the properties they instantiate (e.g., yellowness). The resulting view
is what Siegel calls ‘Radical Naïve Realism’: veridical experience “ . . . consists in a
perceptual relation to a worldly item, and properties are not among the things the
subject is perceptually related to” (2010, p. 358).18
Siegel argues that Radical Naïve Realism is implausible, mainly because our
theory of veridical experience should reflect which properties of a perceived object
make a difference to phenomenal character and which ones don’t (2010, p. 359).
For example, in having a veridical experience of the banana on my desk, the color
of its rind contributes to the phenomenal character of my experience, but the color
of the fruit within does not. Just saying that my experience fundamentally consists
in my perceiving the banana doesn’t capture this obvious phenomenological fact.
It seems that we need to specify which of the banana’s properties I perceive in order
to fully account for the phenomenal character of my experience, contra Radical
Naïve Realism.
17
One might think that o could perceptually appear F to S without it perceptually appearing to
S that o is F on the grounds that only the latter requires that S has the concept of F-ness. However, its
perceptually appearing to S that o is F requires that S has the concept of F-ness only if having experiences
of F-ness requires that S has the concept of F-ness. And if having experiences of F-ness requires that S
has the concept of F-ness, then so does o’s appearing F to S—which means that the claims don’t come
apart after all.
18
Siegel formulates Radical Naïve Realism as a thesis about all non-hallucinatory experiences,
not just veridical experiences. But since I’m concerned only with what the Naïve Realist says about
the latter sort of experience here, I’ve weakened Radical Naïve Realism accordingly. Also, note that if
the Naïve Realist gives up on the claim that we perceive properties, she’ll have to draw the distinction
between veridical and illusory experiences in a different way than I drew it above (a task I’ll leave to
the reader).
Reconciliatory Views
234
While I’m sympathetic to this objection, I can (dimly!) see a way out for the
Radical Naïve Realist. It’s uncontroversial that, when I have a veridical experience
of a banana, my visual system registers some of its features (e.g., the bright yellow-
ness of its rind) but not others (e.g., the yellowish-whiteness of the fruit within).
But it’s not obvious that this uncontroversial fact amounts to perception of some of
the banana’s properties, or to the banana perceptually appearing to be a certain way
to me. For example, one might think that there are subpersonal perceptual states
that carry information about a perceived object’s properties, but the personal-level
upshot of such information processing is a unified experience of an object, as
opposed to one that “carves it up” in terms of its properties (so to speak).19
I’m not confident that this line of thought, when fully spelled out, will vindi-
cate Radical Naïve Realism and the rejection of the argument from appearing it’s
supposed to enable—just as I’m not sure that the denial that experiences are assess-
able for accuracy is a defensible way out of the argument from accuracy. But one
thing I am sure of is that this debate has gotten more complicated than it needs to
be. The driving idea behind these arguments for experiential content is that there
are distinctively perceptual appearances—that we’re failing to capture something
about veridical experience if we confine appearances to the post-perceptual dox-
astic domain. The path of least resistance would be an argument for distinctively
perceptual appearances from a starting point that even the most radical Naïve
Realist would have to accept. Fortunately, I think there is such an argument—let’s
call it the argument from belief generation.
If there are distinctively perceptual appearances, then things perceptually ap-
pear to a subject of an experience to be a certain way (e.g., that there is a yellow
crescent-shaped thing before her), and the proposition that things are that way is
at least necessary for the accuracy of the experience. In other words, if there are
distinctively perceptual appearances, then premise 2 of the argument from accu-
racy and premise 1 of the argument from appearing are true. The argument for
belief generation is essentially a case for the antecedent.
The argument begins with the truism that a given experience naturally gives
rise to particular beliefs rather than others. For example, my experience of a yel-
low crescent-shaped banana naturally gives rise to the belief that there’s a yel-
low crescent-shaped thing before me, but not the belief that there is a purple,
star-shaped thing before me. More precisely, a given experience E is associated
with a particular epistemic appearance that p—the subject is disposed to believe
that p solely on the basis of E, given that she is rational and doesn’t suspect that
E is misleading. Since I am rational and have no suspicions that my experience
of a yellow crescent-shaped banana is misleading, I am disposed to believe (and
indeed do believe) that there is a yellow crescent-shaped thing before me solely on
the basis of my experience.
19
I suspect that something like this line of thought is behind Brewer’s Object View (see his 2008,
especially pp. 171–172).
Experiential Content and Naïve Realism 235
At this point, one should wonder: what grounds the association between E
(say, my experience of a yellow crescent-shaped banana) and the epistemic appear-
ance that p (say, that there is a yellow crescent-shaped thing before me)? What does
my experience have to do with proposition that there is a yellow crescent-shaped
thing before me? How do we get from the former to the latter? A plausible answer
is that it perceptually appears to me that p—i.e., that the epistemic appearance asso-
ciated with E is just the proposition specified by how things perceptually appear to
me. And from here we can establish the Mild Content View. To summarize:
I submit that this argument for the Mild Content View is more dialectically effective
than the arguments from accuracy and appearing. For while the Naïve Realist might
be willing to deny that experiences are assessable for accuracy, or that things per-
ceptually appear to a subject to be a certain way, it would be sheer madness to deny
that a given experience naturally gives rise to particular beliefs about one’s surround-
ings—e.g., to deny that a veridical experience of a yellow banana naturally gives
rise to the belief that there is a yellow crescent-shaped thing before one. (Indeed,
recall that the Naïve Realist had to appeal to epistemic appearances associated with
experiences in order to render the denial of perceptual appearances even remotely
plausible.)
So the argument’s starting point is non-negotiable. But what about its second
premise? There are two broad ways of resisting it: one could put forward an equally
good alternative explanation of the association between experiences and epistemic
appearances, or one could deny that this association requires psychological expla-
nation. As for the first option: the task is to explain, e.g., why the experience I’m
having now gives rise to the epistemic appearance that there is a yellow crescent-
shaped thing before me—in short, the relationship between this proposition and
my perceptual experience. I have no idea what this relationship would be if not
that of the former being the way things perceptually appear to me in virtue of hav-
ing the latter. Any other candidate that comes to mind is more controversial—e.g.,
the proposition being the way I perceptually represent things as being in virtue of
having the experience (as on the Medium Content View). So the proposed expla-
nation seems to be the least we can get away with.
The more promising option is to challenge whether we even need a psycho-
logical explanation of the connection in the first place. One might think that it’s
just a brute psychological fact that the experience I’m having right now gives rise
Reconciliatory Views
236
20
To be fair, one might hold that although we don’t need an explanation of the epistemological
role of perceptual experience in terms of its metaphysical structure, we do need such an explanation
of its phenomenal character and/or its role in facilitating action. However, if one wants explanations
of the latter in terms of the metaphysical structure of perceptual experience, why wouldn’t one want
analogous explanations of the former? The burden is on the proponent of such a view to explain why
the epistemological role of experience differs from the other features in this respect.
Experiential Content and Naïve Realism 237
Naïve Realist could escape the argument from belief generation with her credi-
bility intact. So if Naïve Realism is incompatible with the Mild Content View, then
so much the worse for Naïve Realism. Fortunately for the Naïve Realist, the ante-
cedent is false—establishing this is the first task of the next section.
The conclusion of the argument from belief generation is that the Mild Content
View is true—that for any given experience E, there is a proposition associated
with it that specifies how things perceptually appear to the subject in virtue of
having E. In section 2, I outlined the following argument for the incompatibility of
Naïve Realism with the Mild Content View:
1. Suppose (for the sake of reductio) that veridical experiences have contents
(in the Mild sense that there are propositions that specify the way things
perceptually appear to the subject).
2. Naïve Realism is not explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (a
commitment of Naïve Realism).
3. The fact that a veridical experience has a given content is sufficient to
explain its epistemological, behavioral, and phenomenal features.
4. If veridical experiences have contents, Naïve Realism is explanatorily
redundant with respect to a veridical experience’s epistemological,
behavioral, and phenomenal features (from 3).
5. Naïve Realism is explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (from 1
and 4)
6. Contradiction (2 and 5); veridical experiences don’t have contents.
If Naïve Realism is to be reconciled with the Mild Content View, this argument
must be unsound. I submit that the culprit is premise 3.
To see this, consider the Mild Content View in isolation. All it says is that,
for any given experience, there’s a proposition associated with it that specifies how
things perceptually appear to the subject in virtue of having it. But to say only that
is to leave an important question unanswered: what makes it the case that that prop-
osition specifies how things perceptually appear to the subject? For example, the
proposition that there is a yellow crescent-shaped thing before me specifies how
things perceptually appear to me right now. How did this proposition end up
being cast in this role? Why does it perceptually appear to me that there is a yellow
crescent-shaped thing before me, as opposed to a red, round thing? Since the Mild
Content View doesn’t yield an answer to such questions, the fact that a veridical
experience has a given content in the Mild sense isn’t sufficient on its own to explain
Reconciliatory Views
238
its epistemological, behavioral, and phenomenal features. The Mild Content View
has to be supplemented with further claims that will yield an answer.
And Naïve Realism can come to the rescue. As I previously characterized it,
Naïve Realism is the view that veridical experience fundamentally consists in per-
ception of things in one’s environment. Some Naïve Realists (such as Johnston
2006 and Fish 2009) take this to amount to perception of facts or truthmakers.
Roughly, these are entities constituted by things and their properties, entities of
the form o’s being F (e.g., this banana’s being yellow, Mark’s being to the left of
Bill). As the second label suggests, such entities are what make propositions true.
This banana’s being yellow and before me makes it true that there is a yellow thing
before me; Mark’s being to the left of Bill makes it true that Mark is to the left of
Bill.
Fleshing out Naïve Realism in terms of perception of truthmakers affords a
handy explanation of why things perceptually appear to be the way they do in the
case of veridical experience. We can say that the proposition associated with an
experience that specifies how things perceptually appear to its subject is the one
such that the truthmakers perceived are necessary and sufficient for its truth. For ex-
ample, I perceive this banana’s being yellow crescent-shaped, and to my left, and this
truthmaker is necessary and sufficient for the truth of the proposition that there is
a yellow crescent-shaped thing to my left. It perceptually appears to me that there
is a yellow crescent-shaped thing to my left because the truthmaker I perceive in
the course of my current experience is necessary and sufficient for the truth of this
proposition. In short, the proposal is that the truthmakers the subject perceives
determine which proposition is the content of her experience (in the Mild sense).21
So we should reject premise (3) of the argument for the incompatibility of
Naïve Realism and the Mild Content View. The latter is insufficient to account for
the epistemological, behavioral, and phenomenal features of a veridical experi-
ence, because it is silent on the facts in virtue of which things perceptually appear
to a veridically perceiving subject as they do. Furthermore, Naïve Realism has a
story to tell here—the Naïve Realist can offer an account of the content of a verid-
ical experience in terms of the truthmakers the subject perceives in the course of
having it.
21
The “necessary and sufficient” restriction is required to avoid unwelcome consequences like the
following (respectively): that it perceptually appears to me that either there is a yellow crescent-shaped
thing to my left or 2 + 2 = 4, and that it perceptually appears to me that Heather’s banana is yellow.
This way of specifying the content of veridical experience has a controversial consequence.
The truthmaker this banana’s being yellow and before me is necessary and sufficient for the truth of
the proposition that there is a banana before me. And, as I briefly mentioned at the end of section
9.1, it’s controversial whether the content of experience takes a stand on matters like whether there
are bananas before one. But there is a formulation that is neutral on this issue, namely: the content
of a veridical experience is a proposition concerning which perceptible properties are instantiated in
the subject’s environment, which is such that the perceived truthmakers are necessary and sufficient
for its truth. If it turns out that the property of being a banana isn’t a perceptible property, then this
formulation excludes it from the content of experience.
Experiential Content and Naïve Realism 239
One can use broadly the same kind of reasoning to reconcile Naïve Realism
with the Medium Content View. Recall that the Medium Content View is the Mild
Content View plus the claim that perceptual experience consists in the subject
perceptually representing her environment as being a certain way. On this view,
perceptual representation isn’t what experience fundamentally consists in—this is
what differentiates the Medium and the Spicy Content Views. Since the Medium
Content View is silent on what experience fundamentally consists in, it cannot
provide an exhaustive personal-level psychological explanation of the epistemo-
logical, behavioral, and phenomenological features of an experience. We’re left
without an account of what experience fundamentally consists in, if not percep-
tual representation.
As in the previous case, Naïve Realism can be wheeled in at this point. Notice
that Naïve Realism and the Medium Content View aren’t competitors: the former
is an account of the fundamental nature of veridical experience, while the latter is
not. Hence, it’s possible (at least in principle) to combine them. For example, we
can say that the subject of a veridical experience perceptually represents her envi-
ronment as being a certain way in virtue of perceiving things in her environment—
e.g., that I perceptually represent the proposition that there is a yellow thing before
me in virtue of perceiving the banana’s being yellow. The idea is that in some sense
(which I won’t attempt to spell out here) the subject perceiving things in her envi-
ronment is more basic than the representational state; something about the latter is
explained in terms of the former (see Logue 2013a). Alternatively, we can say that
the representational state is a constituent of the subject perceiving things in her
environment—e.g., that my perceptually representing the proposition that there is
a yellow thing before me is a constituent of my perceiving the banana’s being yel-
low. The idea here is that the representational state is but one part of what veridical
experience fundamentally consists in (see Logue 2012a).22
What about Naïve Realism and the Spicy Content View? Surely, one might
think, the claim that veridical experience fundamentally consists in the subject
perceptually representing her environment as being a certain way renders Naïve
Realism explanatorily redundant. When it comes to veridical experience, Naïve
Realism and the Spicy Content View seem to be competitors—how could verid-
ical experience fundamentally consist in perceptual representation and perceiving
things in one’s environment?
As it happens, this rhetorical question has an answer. Recall that a philosoph-
ical theory of perceptual experience has several explanatory tasks: in particular,
it’s supposed to explain the epistemological, behavioral, and phenomenological
aspects of experience. Thus, it’s in principle possible to divide the labor across Naïve
Realism and the Spicy Content View: say, the latter explains the epistemological
22
Of course, a proponent of the Naïve Realism/Medium Content View package would have to
tell a different story about what non-veridical experiences fundamentally consist in. I’ll set this issue
aside since veridical experience is the main focus of this chapter, but see Logue 2012a for a suggestion.
Reconciliatory Views
240
role of experience while the former yields an account of the phenomenal character
of experience and the role it plays in facilitating action.23 The upshot is that Naïve
Realism and the Spicy Content View need not be in competition with each other.
The fundamental structure of veridical experience could be a composite of a prop-
ositional attitude and a perceptual relation, and it could be that this is the best way
to explain everything that needs explaining.24
Note that I’ve merely offered a template for reconciliation of Naïve Realism
with the stronger Content Views. I’ve said nothing about the benefits we would get
(if any) from combining Naïve Realism with either of them. My aim in this chapter
is simply to show that Naïve Realism is in principle compatible with these views.
Whether either of these combinations is well-motivated is a question that must be
left to another paper.
9.5. Conclusion
Naïve Realists have continued to resist the claim that experience has content, de-
spite compelling arguments in its favor (the arguments from accuracy and appear-
ing). I have proposed a two-prong strategy for talking the Naïve Realist down from
the ledge. First, I offered an argument for the Mild Content View that even the
most radical Naïve Realist wouldn’t reject. Second, I reconstructed what seems to
be the primary argument for the incompatibility of Naïve Realism and experiential
content, and identified a flaw in it. In particular, regardless of which of the Content
Views a Naïve Realist adopts, a story about why a crucial premise in the incom-
patibility argument is false is at least in principle available to her. Hence, Naïve
Realism can be reconciled with experiential content after all.
References
23
One could strengthen the Spicy Content View into the claim that perceptual experience
fundamentally consists in the subject perceptually representing her environment as being a certain
way, and nothing else. Of course, this type of content view is incompatible with Naïve Realism, but in
an uninteresting way (viz., by definition). Thanks to an anonymous referee for pointing out the need
to mention this.
24
I take it that the account of experience defended in Schellenberg 2011 is one version of this
view.
Experiential Content and Naïve Realism 241
Burge, T. (1991). Vision and intentional content. In E. LePore & R. van Gulick (Eds.), John
Searle and his Critics (pp. 195–214). Oxford: Blackwell.
Byrne, A. (2001). Intentionalism defended. Philosophical Review, 110, 119–240.
Campbell, J. (2002). Reference and Consciousness. Oxford: Oxford University Press.
Chalmers, D. (2006). Perception and the fall from Eden. In T. Gendler & J. Hawthorne
(Eds.), Perceptual Experience (pp. 49–125). Oxford: Oxford University Press.
Crane, T. (2009). Is perception a propositional attitude? Philosophical Quarterly, 59,
452–469.
Fish, W. (2009). Perception, Hallucination, and Illusion. Oxford: Oxford University Press.
Johnston, M. (2006). Better than mere knowledge? The function of sensory awareness. In T.
Gendler & J. Hawthorne (Eds.), Perceptual Experience (pp. 260–290). Oxford: Oxford
University Press.
Kalderon, M. E. (2011). Color illusion. Nous, 45, 751–775.
Logue, H. (2012a). What should the Naïve Realist say about total hallucinations?
Philosophical Perspectives, 26, 173–199.
Logue, H. (2012b). Why Naïve Realism? Proceedings of the Aristotelian Society, 112, 211–237.
Logue, H. (2013a). Good news for the disjunctivist about (one of) the bad cases. Philosophy
and Phenomenological Research, 86, 105–133.
Logue, H. (2013b). Visual experience of natural kind properties: Is there any fact of the
matter? Philosophical Studies, 162, 1–12.
Martin, M. G. F. (2002). The transparency of experience. Mind and Language, 17, 376–425.
Martin, M. G. F. (2004). The limits of self-awareness. Philosophical Studies, 120, 37–89.
Martin, M. G. F. (2006). On being alienated. In T. Gendler & J. Hawthorne (Eds.), Perceptual
Experience (pp. 354–410). Oxford: Oxford University Press.
Pautz, A. (2009). What are the contents of experiences? Philosophical Quarterly, 59, 483–507.
Price, R. (2009). Aspect-switching and visual phenomenal character. Philosophical
Quarterly, 59, 508–518.
Schellenberg, S. (2011). Perceptual content defended. Nous, 45, 714–750.
Siegel, S. (2006). Which properties are represented in perception? In T. Gendler & J.
Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford University
Press.
Siegel, S. (2010). Do visual experiences have contents? In B. Nanay (Ed.), Perceiving the
World (pp. 333–368). Oxford: Oxford University Press.
Siegel, S. (2011). The Contents of Visual Experience. Oxford: Oxford University Press.
Stalnaker, R. (1984). Inquiry. Cambridge, MA: MIT Press.
Travis, C. (2004). The silence of the senses. Mind, 113, 57–94.
Tye, M. (2000). Consciousness, Color, and Content. Cambridge, MA: MIT Press.
10
David Lewis thought so. In his view, “someone sees if and only if the scene before
his eyes causes matching visual experience,” where “visual experience has infor-
mational content about the scene before the eyes, and it matches the scene to the
extent that its content is correct.”1 “Visual experience”? Lewis presupposes that
this is a sort of state that “goes on in the brain” (Lewis 1980, p. 239). And he states
that “the content of the experience is, roughly, the content of the belief it tends to
produce”—more precisely, “only if a certain belief would be produced in almost
every case may we take its content as part of the content of the visual experience”
(p. 240).
To see the relevance of these views for the question of this volume, let us
generalize. Seeing is a kind of “perceiving” or “perception”: other kinds include,
at least, hearing, feeling, smelling, and tasting. Visually experiencing—perhaps—is
a kind of “sensorily experiencing”: if so, other kinds include, at least, auditorily
experiencing, tactually experiencing, olfactorily experiencing, and gustatorily expe-
riencing. A scene before the eyes is, perhaps, a kind of “perceptual surround”: if so,
other kinds include, at least, the sounds around the ears, the impingements in and
on the body, the aromas drawn in through the nose, and the flavors in the mouth.
Thanks first and foremost to Jessica Wilson. Thanks also to, in no particular order, Susanna
Siegel, Berit Brogaard, Geoff Lee, David Chalmers, Nico Silins, Jeff Speaks, Heather Logue, Nate
Charlow, Seth Yalcin, Cian Dorr, Tim Button, Laurie Paul, Shamik Dasgupta, Jessica Moss, Ned Block,
John Morrison, Dominic Alford-Duguid, Luke Roelofs, Adam Murray, David Balcarras, Alex Byrne,
Brie Gertler, Jacob Berger, Casey O’Callaghan, Herman Cappelen, Sebastian Becker, Andrew Sepielli,
and many others. Distinctive recognition is due, finally, to Mohan Matthen: for it is through Mohan’s
agency that the lion’s share of the material in this chapter was in a position to take wing.
1
Actually, he thinks the biconditional is “not far wrong.” First, causal chains come cheap: an
intricately refined replacement for “causes” is required to secure sufficiency—the “veridical
hallucination” of the title is a case Lewis takes to show this. Second, eyes are not required: rather, only
some sort of “optical transducer”—as, Lewis thinks, is shown by the “prosthetic vision” of the title. Our
complaints about Lewis’ view will target issues unaffected by these complications, so that ascribing to
242 him the cruder formulation in the body text buys simplicity without the cost of significant misdirection.
Love in the Time of Cholera 243
2
For more on this, see Hellie 2013.
Reconciliatory Views
244
While attention has content, it cannot fail to “match” the perceptual sur-
round—cannot fail to be “correct.” And moreover, the content is not the
“narrow” sort embraced by Lewis (1994). When an aspect of the per-
ceptual surround is targeted by attention, it is present within the stream
of consciousness: the stream of consciousness is characterized by a dis-
tinctively perceptual sort of assurance, or certainty, of the existence of
the target—has the content, “coded” in a distinctive way, that the target
exists. And, more alluringly still, the “nature” of the target is simply
“revealed,” in a way that leaves (at a certain level) no room for doubt
about what it is like.4 While content in general is used to model error
and ignorance, any mistakes or uncertainty I may make are “wrapped
around” a point of certainty about a minor but substantial matter: my
certainty that this exists as such. This certainty is the condition I am in
thanks to having a certain target of attention.
3
Well, can only have “original” content, as against the “derived” content of written messages, and
as against the “contentful stance” we sometimes take toward bread-baking machines.
4
Attention is in this way therefore similar to the “relational states” proposed by a number of
contemporary direct realists—although many of these theorists paint a view incompatible with the
ascription of content.
Love in the Time of Cholera 245
The remainder of this chapter will flesh out this story in detail.
Doing so, regrettably, will require some terminological innovation. For the
ordinary notion of perception is too protean to be useful in philosophical theory, while
the philosopher’s notion of experience carries doctrinal baggage I reject. So the broad
phenomenon of the chapter will be labeled sensory consciousness (sometimes “senso-
rimotor consciousness”), by which we will mean, roughly, those aspects of a creature’s
conscious life that pertain to its “sense-perceptual” or “sensorimotor” condition.
These sensory aspects of consciousness are phenomenologically distinc-
tive: have a character that is immediately striking upon first-person contemplation
of what it is like to undergo them. This distinctive character is often thought to
involve a sort of presentation within sensorimotor consciousness of ingredients of
the objective world (Martin 2004): things around one; one’s own body; the motor
activity of one’s body in relation to the things around one.
If we set sensory consciousness in its broader phenomenological context, it
is this presentational aspect that uniquely qualifies sensorimotor consciousness
to perform a variety of “rational-psychological” duties: duties of a semantic, epis-
temological, or praxeological sort. These include: advancing ingredients of the
objective world as topics for thought and talk (Snowdon 1992); opening a source
of evidence about the objective world (Hellie 2011); providing a sink for agency
in regard to the objective world. So if conscious life in the objective world makes
any sense, the presentational capacities of sensory consciousness must be secure.
Unfortunately, philosophical challenges to presentation remove this security. For
it can seem that what it is like for one can remain fixed over an interval during which
consciousness becomes “disengaged” from the objective world (Valberg 1992); and it
can seem that the subject’s contribution to what sensory consciousness is like threat-
ens to overwhelm any contribution of what is allegedly presented (Hellie 2010).
This chapter will follow out this dialectic. We turn immediately to a theory of
the structure of sensory consciousness; the phenomenon of presentation can be
clearly located within this structure. We then defend the rational-psychological
necessity of presentation. We conclude with discussion of these philosophical
challenges to the possibility of presentation. A crucial aspect of the discussion
will be recognition of the deep nonobjectivity of consciousness, a notion expanded
upon in the technical appendix.
The theory of sensory consciousness used in this chapter is, in outline, the fol-
lowing. At each moment of a creature’s life, the creature and its environment
Reconciliatory Views
246
Fifth, we elaborate the connections between attention and our other notions.
Attention and presentation are linked: what is (or are) present is exactly what
is (or are) the target of attention. The concept of attention is phenomenological
in nature: known and understood ultimately from within conscious life as the
visage of however one comports oneself such that something becomes present.
Nevertheless, attention involves an admixture of the objective: what is a candi-
date target of Sam’s attention is exhausted by what is a constituent of her sensori-
motor state. Still, while the sensorimotor state is objective, that one turns attention
on some aspect of it is not. Attention is therefore the “porthole” in conscious life
through which the objective world drops its anchor; or perhaps the “lashings”
with which conscious life stays moored to the objective world. (Perhaps there is
some objective “realization” of attention: if so, the risk of terminological confusion
would be reduced by calling it “centering” or “tracking.”)
And sixth and finally, we remark on a range of phenomenological features of
attention. Ordinary discourse recognizes looking at x, feeling x, tasting x, smelling
x, and listening to x; these are all varieties of targeting attention on x. Each of these
varieties itself doubtless comes in still more determinate varieties: for example,
staring and luxuriating. Typically we find attention used in conscious life as an
inextricable part of an activity. Consider reading, chasing down a fly ball, assem-
bling a ship in a bottle, dancing, conversing, searching for one’s keys, analyzing a
piece of music or a wine: one performs such an activity attentively just when one
turns attention to those aspects of the sensorimotor process in which the action
unfolds within the objective world. What it is like is not exhausted by what is the
target of attention: intense focus on a tomato and a passing glance differ phenome-
nologically. Moreover, which activity one is performing attentively seems to make
a distinctive phenomenological contribution: an artist making a final survey of a
painting and a gallery visitor studying that painting might glance over the very
same regions of the painting, but what it is like for them in doing so would differ
dramatically.
To summarize. (A) What it is like for Sam—the distinctive character of her
conscious life—is a nonobjective matter. (B) Aspects of what it is like for her
include (but are not exhausted by) every fact concerning what is present within
conscious life. (C) An entity x is present within Sam’s conscious life (and that x
is so present is part of what it is like for her) at a moment just if Sam then targets
x with attention; and (D1) if she does target x with attention, x is a constituent of
Sam’s sensorimotor state. But not conversely: (D2) most constituents of Sam’s sen-
sorimotor state are not targets of attention. While (E1) Sam’s sensorimotor state is
composed in part of what she perceives or senses, (E2) much remains beyond the
scope of Sam’s sensorimotor state. Finally, (F) Sam’s sensorimotor state is an ingre-
dient of the objective world, so that its constituency is an objective matter.
Distinguishing the objective phenomenon of the sensorimotor from the non-
objective phenomenon of conscious life permits an attractive description of the
following sort of case:
Love in the Time of Cholera 249
Fred’s copy of Being and Time: it is on his bookcase somewhere. But where?
Fred combs every inch of the bookcase furiously, repeatedly, unsuccessfully.
His frustration mounts. Until, at last—there it is. Right in front of Fred’s nose
the whole time, he saw it but did not notice it—a source of great consternation.
Recent literature (Block 2011) draws a conundrum from the following assumptions.
(1) Fred sees whatever is right in front of his nose; (2) if Fred sees something right
in front of his nose, that the thing indeed is right in front of his nose is part of what
it is like for him; (3) what it is like for Fred explains what Fred thinks and does.
The conundrum is drawn out as follows. Being and Time is right in front of
Fred’s nose; so by (1), he sees it; so by (2), that Being and Time is right in front of his
nose is part of what it is like for Fred; so by (3), what Fred thinks and does is made
sense of by the fact that part of what it is like for him is that Being and Time is right
in front of his nose. But it isn’t: if that is part of what it is like for Fred, he should
reach out and grab the book rather than continuing the search.
The literature presents a choice between poverty and excessive wealth: some
deny (1), concluding that what we see is impoverished relative to what we think we
see; others deny (3), concluding that what it is like for one is enriched relative to what
we think it is like for one. But the poverty response loses the distinction between
Fred’s case and a search for something simply unseen: the latter should not provoke
the consternation Fred displays in the example. And the wealth response severs the
evident connection between consciousness and rationality: if the location of Being
and Time is within Fred’s conscious life and yet he acts in a way that (we would have
thought) makes no sense in light of that, the rational role of consciousness is cast
into obscurity.
The theory of this section allows the following story. By (3), the rationality
of Fred’s search depends on what it is like for him. By (B), it makes sense to
assume that Fred’s search is rational just if Being and Time is not present within
conscious life. By (C), the search is rational just until Fred’s attention alights on
the book. By (D1), ending the search is only rational if the book is an aspect of
Fred’s sensorimotor state. So, by (E1), at least when the search ends, Fred sees
the book; but moreover, continuing the search can be rational even if the book is
an aspect of Fred’s sensorimotor state: after all, (D2) means that even if attention
has not alit on the book, the book can nonetheless be an aspect Fred’s sensori-
motor state and therefore (E1) seen—preserving consistency with (1). Wrapping
up, (E2) preserves the consternation-free case in which Being and Time is at
home.
So, by asserting both (D1) and (D2), we drive a wedge between the targets of
attention (and onward to presence, what it is like, and rationality) and the constit-
uents of the sensorimotor state (and onward to what is seen)—and are therefore
in a position to reject (2).
But how do we have the right to (D1) and (D2) simultaneously? The literature
embeds a widespread presupposition that consciousness is objective, involving
Reconciliatory Views
250
Attention to a tomato drops the tomato as an anchor of the objective world within
Sam’s conscious life. To appropriate John McDowell’s vivid metaphor (McDowell
1994), it is this anchoring that distinguishes conscious life from a “frictionless
spinning in the void.” As discussed above, a central aim of the philosophy of per-
ception has long been to secure the apparent friction against the concerns to be
discussed in section 4. In this section, we discuss an explanation of why presenta-
tion is worth the bother.
In a nutshell: without presentation, rational psychology—and probably con-
scious life—in an objective world would be impossible. In outline, a sort of “tran-
scendental argument”: (1) Rational psychology is about picture of the world and
stock of actions, both of them grasped through understanding how they evolve
intelligibly (Stalnaker 1984; Anscombe 1963). In particular, (2) one’s picture of
the world evolves through the accumulation of evidence (Lewis 1973), while one’s
stock of actions evolves through the discharging of plans (Bratman 2000); where,
still more specifically (in a way apparently required by embodiment), (3) one accu-
mulates evidence by gradually making more precise one’s certainties about the ob-
jective world as regards the evolving sensorimotor processes of a certain creature
and discharges one’s plans by contouring those same sensorimotor process. So un-
less the link in (3) is intelligible, (RD) our understanding of rational psychology is
thoroughgoingly “semantically defective.” But intelligibility is a phenomenological
notion, in at least the weak sense that (4W) whether someone evolves intelligibly
over an interval is determined by what it is like to be them over the interval; and
perhaps also in the strong sense that (4S) there is nothing to what it is like beyond
that which is relevant to intelligibility. So, by (4W), unless (L) some aspect of what
Love in the Time of Cholera 251
it is like makes the link in (3) intelligible, (RD) follows; and perhaps by (4S), unless
(L), (CD) our understanding of consciousness is also thoroughgoingly semanti-
cally defective. But (RD) is perilously close to the baffling claim that there are no
truths of rational psychology, and (CD) is perilously close to the absurd claim that
there are no truths about consciousness. Fortunately, we can avoid (RD) and (CD),
for (5) presentation within conscious life can, and can alone, suffice for (L)—can
be the aspect of what it is like that makes the link in (3) intelligible.
Now in a bit more detail. Principles (1) and (2) are ancient framework doc-
trines best explored more deeply in another forum. Principle (3) is obvious: for
each of us, there is a creature about which we care in a manner that is absolutely
sui generis, the death of which would extinguish consciousness; it is this creature’s
sensorimotor processes which serve as evidence source and agency sink.
Now to (4W). Intelligible evolution is one with the availability of rationaliz-
ing explanation; of answers to “why” asked with a distinctively rationalizing spirit
(Anscombe 1963). For example: suppose that Fred has leapt to his feet, and that we
wonder why. An “efficient-causal” explanation of the sort offered by physiology is
not what we want: we don’t know any physiology, so such an explanation would be
so much gibberish to us. What we wonder, rather, is what Fred saw in leaping to his
feet at that moment. Citing facts utterly beyond Fred’s ken would therefore be of
no assistance: if Fred thereby narrowly avoided being struck by a flying bottle, that
would be of no explanatory force unless that he did so was part of his picture of the
world. Nor would some sort of “intentional stance”-type story in which some part
of Fred’s brain is treated as performing a calculation on representations (Burge
2005): an explanation offered in the course of Chomsky-type syntactic research
may elucidate how it comes about that a sentence strikes one as structured in this
way rather than that, but it offers no insight into what one sees in being struck by
the sentence in this way rather than that (indeed, one sees nothing in doing so: one
is simply so struck). Instead, what we want to know was what it was like for Fred
in the interval during which he leapt to his feet: what his conscious picture of the
world and conscious aims were such that leaping to his feet was the best action in
his repertoire for achieving those aims in a world like that. If we are told that, in
Fred’s view, the Mayor had just entered the room, that Fred seeks always to obey
protocol, and that Fred’s conception of how to do obey protocol when a high po-
litical figure enters a room calls for leaping to one’s feet, this gives us a sense of
what it was like for Fred; and we do find that if this is what it was like for us, we too
would leap to our feet. This may not be what it was like for us: we think it wasn’t
the mayor, are not especially concerned to obey protocol regarding this mayor, and
think the protocol for mayors doesn’t require leaping to one’s feet anyway. So that
we did not leap to our feet was overdetermined. Nevertheless, when we sympa-
thize with Fred, we understand why he did so.
Now, somewhat more speculatively, to (4S). This principle is in the spirit of,
and inherits the plausibility of, the widely discussed doctrine of “representation-
alism” (Harman 1990b; Chalmers 2004). Separated from its focus on exclusively
Reconciliatory Views
252
At the other pole, for Sam’s conscious life to privilege no objective particular
would be for Sam to be forever ignorant about which location in the objective
world is hers (the situation of the “two gods” in Lewis 1979). Sam’s objective evi-
dence would, she would recognize, update in accord with the peregrinations of
creature Sam; and Sam’s plans would, she would recognize, discharge in a way
mirrored exactly by the motor behavior of creature Sam. Evidence would come out
of nowhere; plans would discharge into nothing. Sam would find herself simply
“saddled” with the evidence, would find her plans simply “falling away”: nothing
in conscious life would be present to make sense of all this coming and going. It
would be cold comfort to superadd the certainty that some creature is, bizarrely,
comporting itself exactly as if a source of Sam’s evidence and sink for her agency.
10.4.1. DISENGAGEMENT
In cases of “illusion” and “hallucination,” one does not recognize that one has
become somehow “out of touch” with one’s environment. Such cases have been
widely thought to demonstrate the impossibility of presentation, on grounds like
the following (Martin 2004):
(1) If anything is ever presented, what it is (perhaps in particular, perhaps
in kind) is then part of the character of conscious life. (2) For any course of
ordinary waking life, one could dream in such a way that what the course of
dreaming is like is no different from what the course of ordinary waking life
is like. Of course (3) the character of conscious life is just exhausted by what
it is like, so that (appealing to (2)) (4) the conscious lives of the ordinary sub-
ject and the dreamer do not differ in character. And if so (appealing to (1)),
(5) whatever is presented to either is presented to both. But surely (6) nothing
(either in particular or in kind) need be presented to both subjects. And if not
(appealing to (5)), (7) nothing is presented to either.
The premisses are (1)–(3) and (6). We have defended (1) and rejected (7) in the
previous section, and advocated (3) in the first section. Little light would emanate
from a challenge to the validity of the argument. So we face a choice of challenging
(2) and challenging (6).
Challenging (6) is the way of Russell (1910–11): even if nothing familiar is
presented to both, perhaps something unfamiliar is—an “internal” sort of
“sense-datum.” A first problem. The unfamiliarity of these internal sense-data is
a perpetual source of widespread philosophical distaste for this approach. This
distaste is a sure sign that we think that what is presented in ordinary waking life
is almost always something “external”; so Russell’s view means that, in ordinary
waking life (and also when we are taken in by a dream), we are almost always
Reconciliatory Views
254
mistaken about what is presented. The alternative this chapter will press predicts
that mistake is (strictly more) rare. In this respect, Russell’s approach is strictly
worse. This leads to a second problem: the stating of the approach undermines its
own solution; so the approach embeds a sort of “pragmatic contradiction.” For if
we come to understand what sense-data are, we will find it easy enough to concoct
a scenario in which what it is like is preserved while the availability of sense-data
is scrambled—destroying the solution. Any difficulty imagining such a scenario is
rooted in our cluelessness about what sense-data are. But that is also bad: for it is
cluelessness about what the theory says. This suggests a third, and fundamental,
problem. Russell’s proposed solution is, ultimately, utterly superficial: the dialec-
tical tension is rooted not in details of the constitution of the objective world, but
rather in our capacity to dissociate the first—and third-person perspectives on
our own lives. (From the first-person point of view, Sam regards herself as seeing
an anteater; when she later learns she was only dreaming, her sense of the situa-
tion from the third-person no longer coheres with her sense from the first-person;
bringing this sense into the “here and now,” one may temporarily adopt an alien-
ated perspective regarding one’s ordinary view as mistaken.) Pushed to its limit,
this strategy leads to Cartesian skeptical hypotheses: and presumably an evil
genius could blast out of the picture any sense-data (or other unfamiliar objective
things) that might have been lying around.
So we need an alternative to (2). Fortunately, the alternative is obvious
(Snowdon 1980–1; Martin 2004; Hellie 2011): when one is taken in by a dream of
looking at a tomato, one mistakenly thinks that one is looking at a tomato; and so
because looking at a tomato is the sort of activity that can be part of what it is like
for one, one therein mistakenly thinks that part of what it is like for one is that one
is looking at a tomato. If one is mistaken in this, then it is no part of what it is like
for one that one is looking at a tomato. And more generally, there is no course of
ordinary waking life for which there is a dream such that what they are like is the
same (though one is always at risk of being taken in).
One might well wonder what it could possibly be to be mistaken about what
it is like. The apparatus of the previous sections sheds light on the phenomenon;
we will develop the position in the course of the following dialogue with a skeptic
about such mistakes (“S” for skeptic; “U” for us):
S. One can’t be mistaken about what it is like for one: one always knows
exactly that it is just like this.
U. Even if so, this infallible but inarticulate sense for what it is like is
irrelevant. For (2) is about cross-comparisons; so unless (2) begs the
question, these must involve some “interpretive grain” mixed in with the
bare “this.” Where there is interpretation, there is misinterpretation. And
where there is misinterpretation, there is a mistaken picture of the world.
S. What is the mistake? (And what is it like when we don’t make the
mistake?)
Love in the Time of Cholera 255
(What it is like when we don’t make the mistake is what lucid dreaming
is like.)
Having made this distinction, the mistake in the bad case is just
that one is seeing and that therefore the presented entity is a physical
object rather than a mental figment. But because any mental figment is
presumably essentially that way and any physical object is presumably
essentially that way, the bad-case judgment this is a physical object is
counteressential and therefore fails to draw up a coherent picture of
the world. The bad case is a case of a bad, because false, presupposi-
tion:5 nothing in the mind is “intrinsically bad”; badness is a Frege case,
resulting from unobvious misalignment of various components of one’s
picture of the world.
S. Are we allowing picture of the world to fix what it is like? If so, I applaud.
(A) One can’t be mistaken about one’s picture of the world, and
(B) presentation seems to have been cut out of the story—so what it is
like is fixed by a trouble-free “narrow” feature about which one can’t
be mistaken (compare the doctrine of “representationalism,” discussed
above).
U. (A) is mistaken. To see this, recall that David Lewis once believed that
Nassau Street runs roughly (to within 10 degrees) north–south, the train
runs roughly east–west, and the two are roughly parallel (Lewis 1982): his
picture of the world was inconsistent. Obviously he spent some time
unaware of this: when he recognized the inconsistency, straightaway it
vanished.
5
Thanks to Dominic Alford-Duguid and Michael Arsenault for making this point especially
sharply.
Reconciliatory Views
256
S. But why does one judge, in the bad case, that the presented figment
is (for example) purple rather than red? In the suprabad case, one
recognizes it to be neither, but simply possessed of a certain feature P
characteristic of some mental figments but not others. What is it about
feature P that combines with the mistaken belief that one is seeing which
misleads us into judging the figment to be purple (Speaks 2013; Logue
2013)? Internalists have a single feature to which they can appeal in
rationalizing both the good and bad case judgments—what about you?
U. The discussion of (6) shows that this is a problem for everyone; what
everyone should say is this. The incoherent bad-case subject has to make
some judgment, of course. But that subject is in the incoherent position of
accepting a counteressential content that this figment is a physical object;
and rational psychology is paralyzed in the presence of incoherence, and
therefore not up to the task of saying which thing it is. So the question
is misplaced. The best we can do is attempt to enter sympathetically
into the position of the other, shut off much of what we know, and think
from a position of self-imposed artificial ignorance what we would do
in the situation of the other. Turning a phrase from McDowell (1994) on
its head: we may have wanted justification, but we will have to settle for
exculpation.
a round tower that looks square from far off, over the large appearance to the mite
of what we find small. It sets in motion the following aporia:
Let Flip have normal color vision for a member of his species. It would
be hopelessly parochial to assert that Norma has but Flip lacks the cor-
rect view on which color this patch of grass is: the cosmopolitan recog-
nizes that each thinks it is green; more generally, that typically everyone
is right about the colors of objects.
But while Norma thinks green is a “cool” color, Flip thinks green is a
“warm” color.6 The higher-order features of “coolness” and “warmth”
are incompatible, cannot be both possessed by a single color; so at least
one of Norma and Flip is wrong. Which is it? It would be hopelessly
parochial to vote for Norma: the cosmopolitan recognizes that both of
them are wrong; more generally, that everyone is mistaken about the
natures of the colors.
But it is within presentation with colors that we arrive at our view on
the natures of the colors. Mistakes about nature are necessary false-
hoods: matters about which we can only be uncertain through gross
confusion. So here again we find a contra-essential error. Whichever
one of them is wrong is not just mistaken: their conception of green
is semantically defective. (Relativization—cool for us, warm for them—
would only push error to the third order: at the third order, warmth and
coolness are presented as “absolute.”)
So presentation is a source of confusion so gross as to engender seman-
tic defectiveness. But the central theoretical role of presentation is as
a source of infallible certainty about objects and their features. Once
burned, twice shy: our misadventures at the higher order should under-
mine our confidence at the lower order—indeed, we now see that they
infect what is meant at the lower order, rendering it unintelligible.
So nothing can fill the theoretical role of presentation: the phenomenon
does not exist. But, as discussed in section 3, without presentation, it
may be that there are no truths about consciousness. So if we reach this
stage, consciousness vanishes.
Perhaps the time is right to revive Kant’s approach to this aporia. The second leap
toward cosmopolitanism—recognition that the structure we find in the world,
we put in in the first place—is both compulsory and forbidden. Compulsory
for the theorist seeking the most objective possible viewpoint. But forbidden
because from that viewpoint, conscious life itself vanishes—and with it, the
theorist.
6
“Cool” and “warm” are a pedagogically-convenient stand-in for whatever higher-order features
we in fact find to distinguish green as we see it from red as we see it.
Reconciliatory Views
258
The second leap toward cosmopolitanism may be one we only perform some-
times, in safe circumstances quarantined from others where it would be genuinely
damaging. Why think either universal cosmopolitanism or universal parochialism
is required of us? A third option is the cynical adventitious cosmopolitanism of
the savvy politician: in the home province, affirming wholeheartedly local paro-
chial biases; in the capital, as easily abandoning them in a cosmopolitan spirit of
national compromise. Diachronically inconsistent; less than wholeheartedly sin-
cere at any moment; quite possibly distasteful if not vicious. And yet life goes on.
This is the First Critique’s discomfiting alternation between transcendental
idealism and empirical realism. Our reasoning tells us that transcendental ide-
alism must be correct. But because what transcendental idealism means can only
be grasped from a viewpoint that is unattainable if ordinary life is to continue, our
desire to leap beyond empirical realism will be forever frustrated.
the structure of the brutely nonconscious, considered in isolation from any pecu-
liarities of consciousness. Are any questions deeply nonobjective—on the one hand
situated; but on the other hand superficial?
Yes. One such question is “is it like this:?” The question is situated. For Fred,
who falsely believes that goats eat cans, the answer is affirmative. For me, it is
negative. Before I learned that goats do not eat cans, the answer was also negative
(but so was the answer to “is it like this:?”, a question I now answer in the affirma-
tive). But the question is also superficial: because it is like that for Fred, I can’t
make sense of what things are like for him if he answers otherwise; and therefore
I can’t make sense of what things are like for him if he takes both answers seriously.
A sentence ϕ entails a sentence ψ just if whenever a context c affirms ϕ,
c affirms ψ . A proposition P is a subset of W, “modal space,” the set of possi-
ble worlds. In a context-sensitive propositional semantics, a declarative sentence
ϕ receives a proposition ϕ as its semantic value against a context c. In a
c
lines of “certainly.”
These rigidifiers exhibit inferential properties corresponding to their associated
notions of affirmation. Verification is “self-dual”: c ⊭ ϕ just if c |= ¬ϕ ; but support
is “non-self-dual”: sometimes both c ϕ and c ¬ϕ . As a result, while A inter-
acts with the classical connectives only classically, ∇ exhibits surprising interac-
tions with the classical connectives. In particular, both dilemma and reductio are
mindset-invalid: although ∇φ ⊣⊢ φ, ¬∇ϕ ¬ϕ and ϕ ∨ ψ ∇ϕ ∨ ∇ψ .
This makes ∇ distinctively useful in representing “transparent certainty”
(Hellie 2011). The equivalence of ϕ to ∇ϕ makes for transparency. And yet while
truth is bivalent, certainty is trivalent: whenever whether-ϕ can be reasonably
asked, it is coherent to accept any of ∇ϕ , ∇¬ϕ , and ¬∇ϕ ∧ ¬∇¬ϕ .
Because its being like such-and-such is a kind of transparent certainty, this
in turn makes ∇ϕ distinctively useful for representing “it’s like this: ϕ” For ϕ and
“it’s like this: ϕ” are equivalent—accepting one is accepting the other. And yet,
while accepting ¬ϕ requires accepting “it’s not like this: ϕ,” the converse is not
so: when consciousness is “grey” regarding whether ϕ, one accepts “it’s not like
this: ϕ” but does not accept ¬ϕ . If the context against which sentences are evalu-
ated represents what it is like for the subject under consideration, we can then say
that c ⊩ ∇ϕ just if, for that subject, it is like this: ϕ.
Reconciliatory Views
260
References
Lewis, David. (1979). Attitudes de dicto and de se. The Philosophical Review, 88, 513–543.
Lewis, David. (1980). Veridical hallucination and prosthetic vision. Australasian Journal of
Philosophy, 58, 239–249.
Lewis, David. (1982). Logic for equivocators. Noûs, 16, 431–441.
Lewis, David. (1994). Reduction of mind. In Samuel Guttenplan (Ed.), Companion to the
Philosophy of Mind (pp. 412–31). Oxford: Blackwell.
Logue, Heather (2013). But where is a hallucinator's perceptual justification? In Richard
Brown (Ed.), Consciousness Inside and Out: Phenomenology, Neuroscience, and the
Nature of Experience (pp. 155–62). Dordrecht: Springer.
Martin, M. G. F. (2000). Beyond dispute: Sense-data, intentionality, and the mind-body
problem. In Tim Crane & Sarah Patterson (Eds.), History of the Mind-Body Problem
(pp. 195–231). London: Routledge.
Martin, M. G. F. (2004). The limits of self-awareness: Disjunctivism and indiscriminability.
Philosophical Studies, 120, 37–89.
Matthen, Mohan P. (2005). Seeing, Doing, and Knowing: A Philosophical Theory of Sense
Perception. Oxford: Oxford University Press.
McDowell, John (1994). Mind and World. Cambridge, MA: Harvard University Press.
Moore, George Edward. (1903). The refutation of idealism. Mind, 12, 433–453.
Noë, Alva (2006). Experience without the head. In Tamar Szabó Gendler & John Hawthorne
(Eds.), Perceptual Experience (pp. 411–33). Oxford: Oxford University Press.
Price, Henry Habberly (1950). Perception. London: Methuen.
Quine, Willard Van Orman (1960). Word and Object. Cambridge, MA: MIT Press.
Rosenthal, David M. (2005). Consciousness and Mind. Oxford: Oxford University Press.
Russell, Bertrand (1910–11). Knowledge by acquaintance and knowledge by description.
Proceedings of the Aristotelian Society, 11, 108–128.
Shoemaker, Sydney (1991). Qualia and consciousness. Mind, 100, 507–524.
Snowdon, Paul (1980–81). Perception, vision, and causation. Proceedings of the Aristotelian
Society, 81, 175–192.
Snowdon, Paul (1992). How to interpret ‘direct perception’. In Tim Crane (Ed.), The Contents
of Experience (pp. 48–78). Cambridge: Cambridge University Press.
Speaks, Jeffrey (2013). Some thoughts about hallucination, self-representation, and
‘There it is’. In Richard Brown (Ed.), Consciousness Inside and Out: Phenomenology,
Neuroscience, and the Nature of Experience (pp. 147–54). Dordrecht: Springer.
Stalnaker, Robert C. (1984). Inquiry. Cambridge, MA: MIT Press.
Thompson, Evan (2007). Mind in Life: Biology, Phenomenology, and the Sciences of Mind.
Cambridge, MA: Harvard University Press.
Valberg, J. J. (1992). The Puzzle of Experience. Oxford: Clarendon Press.
Veltman, Frank (1996). Defaults in update semantics. Journal of Philosophical Logic, 25,
221–261.
Yalcin, Seth (2007). Epistemic modals. Mind, 116, 983–1026.
PART FOUR
Image Content
Mohan Matthen
It is a lamentable but inescapable fact that the world is not always as it appears to our
senses. On the brighter side, one can normally rely on any given appearance being
true. In short, the senses are reliable, but not perfectly so.
These propositions were, throughout most of history, taken for granted by most
philosophers, who, depending on their philosophical bent, emphasized one of them
over the other. In ancient Greece, Plato led the pessimists. He maintained that the
senses are unreliable and inchoate; what they tell us must be clarified, if not wholly
supplanted, by reason. The ancient sceptics overshot Plato: according to them, the
senses carry no credibility; nothing they tell us is more probable than the opposite.
Aristotle led the optimists: the senses provide a platform for systematic knowledge, he
says (Post. An. II 22). The Stoics went even further—according to them, some kinds
of perception are guaranteed to be true. Philosophers in ancient and medieval India
pursued the same lines of thought. They acknowledged sensory illusion. The central
problem of epistemology, according to them, was whether sensory illusion betokens
a veil of ideas between the perceiving subject and the external world. Philosophy
in the modern period continues these debates, albeit with increasing precision and
refinement.
All of these approaches to perception imply (on very natural assumptions) that
the perceiving subject is presented with a proposition—the proposition that the world
is a certain way. This proposition is often entitled the content of a perceptual state.
A subject’s perceptual experience is true or false according to whether the proposition
it presents—that is, the proposition that constitutes its content—is true or false. When
something looks green, the perceiver is being presented with the proposition that it is
green; if the thing is green, the propositional content of this presentation is true, and
hence the appearance itself is also true.
As the present volume demonstrates, claims such as these have come under
fire from many directions. Even in the ancient world, Epicurus denied that sensory
appearance could be false. His argument is especially interesting to my project in this
chapter because it explicitly rests on something like the idea that perception is an
image. “The portrait-like resemblance of impressions . . . would not exist if the things
265
Imagistic and Possible-Word Content
266
which we come into contact were not themselves something,” he wrote.1 The idea is
that external things leave their mark on the senses; these marks are the consequence
of a process of transmission from external object to the senses. Since this process
preserves a certain resemblance, sensory impressions are “portraits.”
If Epicurus is right, the “portrait-like” sensory appearance is merely a causal
trace. As such, it cannot have propositional content. The image cast on a screen
by a slide-projector is a copy of the slide, but it would be strange to call it “true”
for this reason. Epicurus thought that sensory appearance is like this projected
image. We can use it to figure out the condition of external things by inference
from nature of effect to nature of cause. Thus, he holds that “falsehood and error
are always located in the opinion which we add.” In effect, this position conflicts
with the idea that sensory appearance can be true: for if falsehood is “located in the
opinion which we add,” so also is truth. In him, therefore, we find one form of a
“no-content” view of the senses. (The impressed-image theory is inconsistent with
what we now know of sensory processing, as I will indicate in section 11.9, and an
untenable basis for a no-content view.)
These ancient positions, as well as their more recent descendants, are interesting
not just for what they say, but also for what they omit. Sensory content is imag-
istic, as Epicurus explicitly acknowledges. Opinion and belief, by contrast, are (in
standard cases, at least) linguistically expressed. Belief is not normally directed at
an image—my opinion that the sky will be blue today is directed toward a sentence,
not an image of the sky. One can, of course, believe an image, as I will later indicate.
But there is still a serious omission in contemporary theory: there is no discussion
of how an opinion can interact with an image. Since the former is linguistic in the
manner it conveys content, and the second is not, this is a problem—images cannot
bear on sentential attitudes directly; only a sentential description of an image can
do so.
In this chapter, I seek to clarify the nature of image content and its expressive
limitations. In particular, I shall inquire into how the characteristics of image con-
tent bear on the issues of representational content indicated above.
1
Letter to Herodotus, § 51 (= Long and Sedley 15A11). Note: ‘portrait’ = eikon.
2
X perceives p only if p: for ‘perceives’ is what philosophers call “factive”. In my usage, “perceptual
experience” is experience as in perception, i.e., experience involving sensory qualia with the feeling
of here-and-now as per (b). Hallucinatory experience may be perceptual in this usage, though
hallucination is not perception, on account of its failing the factivity condition.
Image Content 267
But this is a characteristic that perceptual experience shares with iconic memory
and sensory imaging. It is different from these other imagistic states because:
Memory and sensory imaging do not present states of affairs as occurring here and
now. This is how they differ from perceptual experience.3
It is hard (as we shall see) to understand how both (a) and (b) could be true.
Experiences of the sort described by (a) are imagistic—some of the key character-
istics of imagistically expressed content are laid out in sections 11.3‒11.4. But, as I’ll
argue in sections 11.6‒11.8, images are incapable of fixing places and times, except
relative to one another. That is, an image can present one thing as further away
and over to the left of another, but not as in a particular place (e.g., London) or at a
particular time (e.g., five minutes ago). So how does perceptual experience present
something as here (i.e., in this particular place) and now (at this time)? How can it?
This is just one question that arises from taking image content seriously.
Before we go on, a clarificatory remark is in order. Often, image content
is contrasted with propositional content, on the grounds that images cannot be
assessed, as propositions are, in terms of truth and falsity and related semantic
evaluations. In my view, this is a mistake.
Here is why. On any non-deflationary account (see section 11.9), an image is an
image of a state of affairs—an image of a way the world can be. (Put aside impossible
images, which require a more complicated account.) For example, an image of a face
presents a state of affairs in which a face exists in a certain place relative to a point
of view—a face that is such-and-such color, surmounted by thinning hair, with a
long nose, . . . and so on. The image may, further, present the face as singing a song,
having soft skin, and possessing other non-visual features. (My notion of an image is
multisensory; the “sensory qualities” of (a) above may belong to any modality.) We
can, therefore, assess the relationship between an image and any possible world. Is
the image an image of that possible world? Or: how true is the image of that world?
And this is just like the relationship between a sentence and a possible world.
An image is an image of a type of situation. It is not an image of a particu-
lar situation—rather, it is an image of any situation of the type that it depicts. In
order for it to be true—that is, true non-relationally, not merely true of a speci-
fied situation—it has to be applied to, or asserted of, a particular situation or ex-
istentially quantified over a range of situations. Perceptual experience applies its
3
Hume distinguished perceptual experience (“impressions”) from memory and imagination by
the “force and liveliness” of the former. He would have said that a state of affairs is experienced as “here
and now” because the idea of the state of affairs is livelier, and is hence an impression (not merely an
idea). It is noteworthy in view of the Shared Content Argument in section 11.6, that Hume defines
perceptual experience in a way that is independent of content.
Imagistic and Possible-Word Content
268
4
Tim Crane (2009, pp. 467–468) argues that pictures cannot be asserted except by using
non-pictorial means (or “symbols,” as he says). Taking a pass on ‘symbol,’ I agree. Crane thinks that this
thesis gives him reason to deny that perception is a propositional attitude. My claim is that perceptual
experience asserts an image of the here-and-now simply by virtue of being the kind of experience it is. It
is for this reason that perceptual experience can be true or false. This is why it is a propositional attitude.
5
In an otherwise excellent discussion of images from which I have learned much, Charles Siewert
(1998) suggests that a merely imagined scene could turn out to be true—for instance, when I imagine a
unicorn and it turns out that there is a unicorn that exactly matches my image. I think this is a mistake.
There was nothing in my imagining that conveyed to me that my image was an image of this or any
other actual creature.
Image Content 269
6
Christopher Peacocke’s (1983, 1992) notion of “scenario content” is restricted to such
feature-location content, though it is less conservative than the empiricist tradition, inasmuch as it
allows for perceptual representations of three-dimensional space. Austen Clark’s (2000) idea of located
features is another example.
7
It was, moreover, a matter of debate among members of the Vienna Circle whether protocol
sentences were descriptions of the subject’s own phenomenal state (and hence private) or of
feature-instances in shared real space. Carnap seems to have started with the phenomenal interpretation
and moved, under the influence of Neurath, toward the public space interpretation (Uebel 1992). I am
taking protocol sentences in the second way.
Imagistic and Possible-Word Content
270
For reasons I won’t discuss here, this is a discredited program. From my point
of view, the main objection is that there is no clear way to capture what an A380
looks like, given that we see such objects from many different points of view in
diverse conditions of illumination (etc.). Even more obviously, we cannot define
how sexual interest appears to the senses, in all of its many different contexts. In
short, we cannot define or “construct” such objects or features in terms of sensory
qualities alone.8 The interesting question arises, therefore, whether one can go di-
rectly from perceptual images to “rich” perceptual beliefs by-passing the interme-
diary stage of protocol perceptual beliefs. The problem is that this would require
an indirect and problematic image-sentence transition.
In sum, the obstacle to a satisfactory account of perceptual belief-formation
is that, putting protocol sentences aside, we don’t have a good general account of
transitions from image-content to sentence-content. On the other hand, we do
have good general accounts of sentence-sentence inference. The logical empiri-
cists solve the problem of belief-formation by relying on the special case where
we do have a clear transition. They assumed that they could then use the method
of construction to carry the inference train forward to rich perceptual beliefs.
Given the failure of their program, it seems that we would want to explore learned
image-sentence transitions. And this takes us back to square one: we have at best
a poor understanding of how we reason with images, as opposed to sentences.
Philosophers who discuss perceptual belief rarely see the need for an account of
this.9 John McDowell (1994) is an exception: his assertion of conceptual content
in perception is presumably meant to serve just this role. However, conceptual
content is just a placeholder, since we have no good idea of how the conceptual
content of perceptual images interact with concepts as they are found in sentences
and language (including the language of thought).
Though I will make a tentative suggestion about the transition from images
to rich perceptual beliefs, my purpose in this chaper is mainly negative. I shall
sketch some important characteristics of sensory images, with a view to showing
how these lead to certain expressive limitations. This rules out certain kinds of per-
ceptual content. More specifically, I’ll argue that images cannot express particular
times and places (though they can express relations of locations and times, such as
‘x is to the left of/earlier than y’). As well, they cannot refer to particular individu-
als. So if perceptual experience and iconic memory refer to particular individuals,
or express fixed spatiotemporal locations, they must do so in virtue of something
outside of the image they present. I’ll make a suggestion as to what this might be.
8
Others object on the grounds that the program is foundationalist. But first, this might be
historically false (Uebel 1993), and secondly, there is nothing wrong with foundationalism.
9
See, however, Elisabeth Camp (2007) for an insightful discussion of “Thinking with Maps.”
Camp uses “icons” with conventional meaning to carry some of the weight of the informational
content of maps; she is concerned with displays in which imagistic content carries only some of the
representational burden. In this chapter, I am concerned with pure image content—no captions,
shading, and other such symbols with conventional meaning.
Image Content 271
Finally, I’ll argue that on natural assumptions about sensory images, defla-
tionary theories of content are on shaky ground: it’s hard to understand images as
not expressing content. On the other hand, it is probably true that images do not
express rich sentential content.
Perceptual experience has image content, but so also do sensory imaging and epi-
sodic memory. In this section and the next, I will lay out some general characteris-
tics of sensory image content as it is found in these types of experience.
10
See Matthen (2004) and (2005, ch. 12) for a fuller defense of the predication claim than is
possible here, as well as an argument against Austen Clark’s (2000) thesis that features are merely placed
in visual field-places rather than predicated of individuals such as material objects. A brief argument is
given at the start of section 11.3 below.
11
Xu (1997) argues that material object is the only sortal at work in visual perception. Her
argument is meant to exclude finer sortals, such as man, but she doesn’t consider shadows and the like
for which the rules of overlap and interpenetration are different.
Imagistic and Possible-Word Content
272
I shall argue that this particularity comes from outside the image. Sensory images
mark places (identified, not absolutely, but relative to a point of viewing) as occupied
by some object of a certain (sortal) type.12 (I’ll argue in support of this in section 11.5.)
To sum this up, the content of a sensory image can be expressed by a set of
what I call “(predicative) feature-placing structures” of the form <sortal S, feature
F, location L>. Each such structure expresses a situation-type in which an object
of sortal type S instantiates feature F occupies location L (identified relative to a
point of view).13 As mentioned earlier, an experience acquires truth-value when
such a feature-placing structure is asserted of a particular situation, or quantified
over a range of situations.
In sensory images, locations are given from some point of view. Some say that
the point of view is that of the subject, but this is not always true. In “observer
perspective” memory images, the subject images herself from a removed perspec-
tive. For example, you might recall yourself going to kindergarten, viewing yourself
from above or from behind. (It is noteworthy here that the self is not identified
by recognition; rather, the self comes “tagged” as such.) As well, one can imagine
how a scene would look from a particular point of view without imagining oneself
placed at that point of view. The decisive examples are those in which you yourself
appear in the scene. For instance, you can imagine what you look like from behind
if you were to cut your hair differently. Such examples demonstrate the possibility
of imaging a scene, even one that does not have oneself as a part, without imagining
that one is oneself the viewer. Of course, this is not true of perceptual experience.
The “here and now” of such experience depends on the point of view being that
of the perceiver. But one should not generalize from this to other types of sensory
experience.
12
My position is that sensory image content is “abstract” in the sense of Tye (2000, p. 62). That
is, it does not involve particular objects. However, I hold that perceptual (and possibly recollective)
experience is object-involving.
13
The complete set of feature-placing structures available to a perceiver at a time would amount
to something like Peacocke’s (1982) scenario content, adding sortals to identify the subjects of feature
attribution, and subtracting the assumption that the point of view is that of the subject.
Image Content 273
yourself and a source of loud music, without “hearing” silence there. But you can
also hear a room go silent, or silence in a particular place. Similarly, vision gives
you no information about occluded places or about the backs of objects you see
(though it represents these objects as having backs)—but it can also reveal that a
particular place is not occupied.
I’ll return to the Predicative Feature-Placing condition in section 11.4, where
I discuss the objects of each sense modality.
14
Traditionally, the visual field was taken to be two-dimensional. For powerful arguments in
favour of a three-dimensional field, see Austen Clark (1996). See also Matthen (2005, ch. 12).
15
Colin McGinn (2004, p. 23) suggests that imagined spatial matrices may have gaps. His point
seems to be that the two towers exist in a single matrix, which has a gap in it. I am not sure what
gaps are in this context, or why they do not have determinate size. This is why I prefer to say that the
matrices are separate. This is not a big point.
16
Interestingly, music perception seems to present a kind of non-spatial motion (Charles
Nussbaum, forthcoming). That is, it is very natural to think of harmonic progressions, rhythmic lines,
etc. as moving. Most likely, this is a cross-modal effect, as is the perception of higher frequency notes
as spatially higher.
Image Content 275
As J. J. Gibson writes:
In active touching and looking the observer reports experiences [that] corre-
spond to the environment instead of to the events at the sensory surface. The
experiences noted with passive stimulation can scarcely be noticed, if at all.
(1962, pp. 489–490)
So touch seems simultaneously to deploy two coordinate systems, though (as the
transition between passive and exploratory touch shows) the two systems repre-
sent the same spatial matrix.
17
Flavor is a very complex case, since its components come from a variety of receptors, some in
the tongue, some in the nose, some in the face. All of these are synthesized into a single experience and
“referred” to the mouth, or to the object in the mouth. (After tastes are in the mouth, but obviously
not in any object.)
18
Peacocke: “A visual perceptual experience enjoyed by someone sitting at a desk may represent
various writing implements and items of furniture as having particular spatial relations to one another
and to the experiencer . . . A sensation of small (sic), by contrast, may have no representational content
of any sort, though of course the sensation will be of a distinctive kind” (5).
19
Batty (2010) distinguishes between smell, which is an olfactory property of the perceiver’s
surroundings, and odor, which is a chemical substance in the air.
Imagistic and Possible-Word Content
276
20
It is common among philosophers to collapse these into the two-dimensional projection of
a shape—they say that a square viewed at an angle looks like a trapezoid—but see Matthen (2010b).
21
Compare Boghossian’s (1994) condition of Transparency: (a) If two of a thinker’s token thoughts
possess the same content, then the thinker must be able to know a priori that they do; and (b) If two of
a thinker’s token thoughts possess distinct contents, then the thinker must be able to know a priori that
they do. I do not endorse this condition in its full generality.
22
The Property Coding condition excludes the kinds of devices suggested by Camp 2007 for
map-like representations: for example, “A black (or other fully-saturated) icon [c]ould represent
certainty that the relevant object/property is at that location” (163). This would entail that black would
represent both blackness and certainty.
Image Content 277
The feature-object structure of perceptual content is, again, relatively obvious for
vision. Here is a simple pair of visualizations that demonstrates it.
First, imagine a row of transparent objects. Imagine that starting from the left
and continuing towards the right, each successively turns blue and then back to
transparent, perhaps by an internal light being turned on and then off. Imagine
that this is done slowly enough that the display does not fuse, and is not seen as
a single moving object. Call this the Shifting Blue phenomenon or SB.
Now imagine just one of these objects in its blue state moving from left to
right. Call this the Moving Object phenomenon or MO.
23
Carey: “I assume that representations are states of the nervous system that have content that
refer to concrete or abstract entities (or even fictional entities), properties, and events” (2009, p. 5).
Imagistic and Possible-Word Content
278
Of course, there is more than one difference between SB and MO: for instance,
there are spatial gaps in SB and a smooth trajectory in MO. However, the phe-
nomenological difference between SB and MO is not limited to the sequence of
blue-occurrences in various places. It includes that a single moving blue object
is presented in MO, but not in SB. (Actually, if SB is speeded up, it will look just
like MO: the gaps will disappear.) This difference between the two presentations
is not captured by all the feature-place pairs in each. For since the idea of move-
ment is one of a single object taking its properties with it, this difference must be
accounted for by positing that blue is predicated of one object in the phenome-
nology of MO, and of many in SB. (Objects need not be material objects: colors
may appear to belong to fringes, shafts of light, shadows, vapors, auras, etc.)
Motion perception (and also perception of change) demonstrates that visual
content represents features as belonging to movable objects, mainly material
objects. What about the other senses?
i. In audition, the smallest objects are sounds. Sounds are events that
cause the waveform vibrations in the air that excites the ears (Casati
and Dokic 1994; Pasnau 1999; O’Callaghan 2009): events such as a bow
being drawn across a string or the vibration of somebody’s vocal chords.
(More specifically, they are the last cause of such waveform vibrations
that are not themselves waveform vibrations: Matthen 2010a, pp. 82–83.)
Auditory qualities—loud, soft, high, low, etc.—are transmitted by waves
in the acoustic medium, and correlate with properties of those waves.
Nevertheless, what we hear is a property of the event that causes the
waveform. Audition attributes features to located events: for instance,
a man is playing the drum loudly there. This attributes a property to the
event that consists of the man beating a drum; it does not attribute a
property to the sound waves that the beaten drum emits.
Audition also identifies composite sounds: melodies, phonemes,
speech streams. These are composed of sounds, but the auditory system
recognizes them as single entities (Matthen 2010a). These objects have
properties that are detected by audition: a falling tone, a diminuendo, a
harmonic resolution, etc. are such properties.
ii. In active touch, the primary objects are surfaces of objects, but when the
subject can either see the object he is touching, or can actively touch it
by stroking, feeling, and palpating, surfaces appear voluminous (Gibson
1962). When vision is active alongside touch, the objects to which tactile
qualities are attributed are often those defined by vision, both with respect
to their location and with respect to their exact contours. (In passive
touch, as discussed earlier, the objects are part of the body.)
iii. The objects of olfaction are odors, distributions of volatile chemicals in
the atmosphere, but they are not, of course, presented as such. Odors
have extended location, though it takes movement to detect this. Odors
Image Content 279
are the bearers of smells, which are olfactory qualities such as sweet
and burnt.24 One can also smell material objects, but this takes active
olfaction: moving around, picking things up, and sniffing. In such cases,
again, the bearers of olfactory qualities are defined by other senses—
vision and touch, in particular.
iv. Finally, flavors are typically attributed to objects in the mouth, which are
located by touch. Analogously to vision, it is possible to have free-floating
flavors, such as after-tastes, located in the mouth, but with flavor too these
are normally not mistaken for flavors that reside in a sensed object.
Though images and sentences both express propositions, there are propositions
that cannot be expressed by images. For example, as I said earlier, negation and
universally quantified propositions cannot be expressed by a sensory image.
Now, some sententially expressed accounts of sensory content seem to violate
the above conditions on image content. For example, Susanna Siegel (2006) claims
that when a subject S is looking at an object o that appears to be external to her, S’s
visual experience has the following content:
(PC) If S substantially changes her perspective on o, her visual phenome-
nology will change as a result of this change. (2006, p. 358)
PC violates every condition on images laid down in the preceding section. For
instance, PC seems to imply that one of the things visual experience presents as
content is its own “visual phenomenology”: that is, according to PC, visual expe-
rience does not merely possess phenomenology, but reveals something about this
phenomenology. But this violates the Property Coding condition because there is
nothing that visual phenomenology looks like. (Rather, visual phenomenology is
how objects and features look.) Again, it is unclear how a visual experience could
represent the “will change if ” conditional in PC. Without wishing to question
that some conditionals are implied by visual states—for example, the conditional
proposition that a moving object will arrive at a particular place if it continues
to move in the direction it is now moving—it is unclear how conditionals can be
visually presented, in particular the conditional stated in PC. And where is this
conditional? Perspective-change has spatiotemporal parts—but how are they rep-
resented? For these reasons, it is hard to see how an image could represent PC.
Elsewhere Siegel says, more simply, that
24
I take the distinction between odors and smells from Clare Batty (2010); however, she argues
that the object of olfaction is a “know-not-what”.
Imagistic and Possible-Word Content
280
The position I have articulated on the expressive limitations of images has inter-
esting historical counterparts. Consider Hume’s argument that existence is not
imagistically represented:
I do not think there are any two distinct impressions, which are insep-
arably conjoin’d. Tho’ certain sensations may at one time be united, we
25
See McGinn 2004 for other such differences between perceptual and non-perceptual sensory
experiences.
Imagistic and Possible-Word Content
282
The given in my experience of, say, looking at a ripe tomato does not contain
judgments such as “That is a tomato,” “That tomato is red,” and “I am seeing a
tomato.” It is plain on reflection that my visual experience, when considered
Imagistic and Possible-Word Content
286
in isolation, does not entitle me to the judgement that the object before me is
a tomato.. .
The second consideration is that experience is passive, and it is always a
good policy not to assign fault to the passive. . . . When I have what is called
a “misleading” experience, experience has done nothing to mislead me. The
fault, if any, lies with me and my beliefs—beliefs for which I am responsible.
When on a foggy day, I take a pillar to be a man, it is not my visual experience
that tells me that there is a man before me; the experience is ill-equipped to do
such a thing. I form the belief that there is a man. (2006, pp. 185–186)
26
Gupta’s view is similar to Charles Travis’ (2004) view. Travis’ argument is considerably more
complex and nuanced, but in the end, I think it too treats sentential expressions of perceptual content
as if they were independent of underlying image content.
Imagistic and Possible-Word Content
288
References
27
I am very grateful to Chris Gauker for detailed comments after a careful reading of the whole
manuscript.
Image Content 289
Gupta, Anil (2006). Experience and knowledge. In T. Gendler & J. Hawthorne (Eds.),
Perceptual Experience (pp. 181–204). Oxford: Clarendon Press.
Johnston, Mark (1992). How to speak of the colors. Philosophical Studies, 68, 221–263.
Lycan, W. (2000). The slighting of smell (with a brief note on the slighting of chemistry).
In N. Bhushan & S. Rosenfeld (Eds.), Of Minds and Molecules: New Philosophical
Perspectives on Chemistry (pp. 273–290). New York: Oxford University Press.
Martin, M. G. F. (2002). The transparency of experience. Mind and Language, 17, 376–425.
Martin, M. G. F. (2006). On being alienated. In T. Gendler & J. Hawthorne (Eds.), Perceptual
Experience (pp. 354–410). Oxford: Clarendon Press.
Martin, M. G. F. (2010). What’s in a look? In B. Nanay (Ed.), Perceiving the World (pp.
160–225). New York: Oxford University Press.
Matthen, Mohan (2004). Features, objects, and places: Reflections on Austen Clark’s Theory
of Sentience. Philosophical Psychology, 17, 497–518.
Matthen, Mohan (2005). Seeing, Doing, and Knowing: A Philosophical Theory of Sense
Perception. Oxford: Clarendon Press.
Matthen, Mohan (2010a). On the diversity of auditory objects. Review of Philosophy and
Psychology, 1, 63–89.
Matthen, Mohan (2010b). How things look. In B. Nanay (Ed.), Perception and the World
(pp. 226–252). New York: Oxford University Press.
Matthen, Mohan (2012). Visual demonstratives. In A. Raftopoulos & P. Machamer (Eds.),
Perception, Realism, and the Problem of Reference (pp. 43–67). Cambridge: Cambridge
University Press.
Matthen, Mohan (forthcoming). Active perception and the representation of space. In D.
Stokes, M. Matthen, & S. Biggs (Eds.), Perception and Its Modalities. New York: Oxford
University Press.
McDowell, John (1994). Mind and World. Cambridge, MA: Harvard University Press.
McGinn, Colin (2004). Mindsight: Image, Dream, Meaning. Cambridge, MA: Harvard
University Press.
Nussbaum, Charles (forthcoming). Musical perception. In M. Matthen (Ed.), The Oxford
Handbook of the Philosophy of Perception. Oxford: Clarendon Press.
O’Callaghan, Casey (2009). Sounds. Oxford: Clarendon Press.
Pasnau, Robert (1999). What is sound? Philosophical Quarterly, 49, 309–324.
Peacocke, C. (1983). Sense and Content: Experience, Thought and their Relations.
New York: Oxford University Press.
Peacocke, Christopher (1992). A Study of Concepts. Cambridge, MA: MIT Press.
Pollock, John (1974). Knowledge and Justification. Princeton: Princeton University Press.
Siegel, Susanna (2006). Subject and object in the contents of visual experience. Philosophical
Review, 115, 355–388.
Siewert, Charles P. (1998). The Significance of Consciousness. Princeton: Princeton University
Press.
Snowdon, Paul (1981). Perception, vision and causation. Proceedings of the Aristotelian
Society, New Series, 81, 175–192.
Snowdon, Paul (1990). The objects of perceptual experience. Proceedings of the Aristotelian
Society, Supplementary Volume 64, 121–166.
Spelke, Elizabeth S. (1990). Principles of object perception. Cognitive Science, 14, 29–56.
Travis, Charles (2004). The silence of the senses. Mind, 113, 57–94.
Imagistic and Possible-Word Content
290
Uebel, Thomas E. (1992). Rational reconstruction as elucidation? Carnap in the early proto-
col sentence debate. Synthese, 93, 107–140.
Uebel, Thomas E. (1993). Neurath’s protocol statements: A naturalistic theory of data and
pragmatic theory of theory acceptance. Philosophy of Science, 60, 587–607.
Xu, Fei (1997). From Lot’s wife to a pillar of salt: Evidence that physical object is a sortal
concept. Mind and Language, 12, 365–392.
Xu, Fei (2007). Sortal concepts, object individuation, and language. Trends in Cognitive
Science, 11, 400–406.
12
Sometimes our senses mislead us. Sometimes things are as they appear. One
natural way to take these anodyne remarks is as indicating that our perceptual
experiences are sometimes inaccurate, sometimes accurate. Perceptual experi-
ences, then, have accuracy conditions. Where there are accuracy conditions,
there is representational content. Historically, the idea that perceptual expe-
riences have representational content was not the usual one. The most pop-
ular view was that content resides further downstream in the beliefs that are
formed on the basis of perceptual experience. Perceptual experiences them-
selves are ‘blind’. Sometimes, this view is cashed out further by holding that
one who undergoes a perceptual experience is presented with an appearance
(or sense-datum); at other times, it is held that undergoing a perceptual expe-
rience is a matter of being appeared to in a certain way. On both these views,
there is no accuracy or inaccuracy at the level of perceptual experience. Thus,
there really are no purely perceptual illusions or hallucinations. When one sees
a blue object, say, that looks green, it is not that one’s experience misrepresents
the object as green, but rather that one is presented with an appearance that is
green. This appearance is the ‘look’ of the object—and the object genuinely has
that look, so there is no inaccuracy (and no need to admit that the experience
has a representational content). If one forms the belief that the object is green,
then mistake arises there: the object is not as the belief represents it to be. But
there is no mistake in the experience itself.
This chapter assumes that such nonrepresentational approaches to percep-
tual experience are mistaken. Views that countenance appearances (or ‘looks’)
as real entities have to face a range of very puzzling questions; and adverbialist
approaches need to construct highly complex and contrived adverbs to account
for the full range of perceptual scenarios. What has motivated these views at least
for some is the thought that there cannot be content without concepts, and so
content only really arises at the level of thought or belief. The present chapter also
assumes that this is a mistake. There is no difficulty in making sense of the idea 291
Imagistic and Possible-Word Content
292
1
In particular, the passage that goes as follows: “When I see a tomato there is much that I can
doubt. I can doubt that it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can
doubt whether there is any material thing there at all. . . . One thing, however, I cannot doubt: that there
is exist a red patch of a round and somewhat bulgy shape” (Price 1932, p. 3).
2
I shall ignore the case of de re hallucinations. These may be regarded as illusions of a special sort.
As to how I am using the expression ‘representational content’ in sections 12.1 and 12.2 below, I take
it that a visual experience v, in having the representational content that p, is related to an entity picked
out by ‘that p’ that is v’s representational content. I further take it that v has accuracy conditions and
that v’s having the accuracy conditions it does follows from its having the representational content it
does. I intend my question, “What is the content of Keith’s experience?” to be asking for an elucidation
of the nature of that content. I should add that I do not hold, as some do, that the content itself has
accuracy conditions, though I cannot pursue this point here. On my view, representational contents
generally do not have accuracy or truth conditions. Accuracy (or truth) and inaccuracy (or falsehood)
are properties of the vehicles having content. Finally, I should note that in section 12.3 below, I present
an alternative notion of content.
What is the Content of a Hallucinatory Experience? 293
suggests that the object itself figures in the content of the experience, assuming
that experience is representational at all.
In cases of illusion, the perceived object appears other than it is. In such cases,
intuitively, as already noted, the perceptual experience is inaccurate.3 And it is so
precisely because the object is not as it appears to be. The simplest explanation
of this, in my view, is that, where there is a perceived object, a perceptual experi-
ence has a content into which the perceived object enters along with its apparent
properties.4
Once it is acknowledged that in veridical and illusory cases the seen object
is a component of the content of the experience and thus that the content itself is
singular, a puzzle arises.5 In standard hallucinatory cases there is no object with
which the subject is in perceptual contact and correlatively no singular content.6
So, again, what is the content of Keith’s experience?
One possible answer is that the content is existential in the hallucinatory case.
Keith’s visual experience represents that there is something red, round and bulgy
before him, and since there is no such thing, his experience is inaccurate. The com-
bination of views that results is unlovely and implausible; for it is forced to postulate
a displeasing and radical asymmetry in cases that pre-theoretically seem alike.
Another possible answer is that Keith’s experience has no content at all in the
hallucinatory case. Keith is simply sensing a red, round, bulgy sense-datum or he
is sensing redly, roundly, and bulgily. Again, the resulting combination of views for
the veridical and hallucinatory cases is unlovely and implausible. And, as already
noted, there are other difficulties for views that introduce sense-data or go adver-
bial. This is so even if these views are restricted to hallucinations.
These reflections suggest that in the hallucinatory case, we should say that
there is content of the very same sort as in the veridical and illusory cases—con-
tent that is just like singular content but with a gap or hole in it where the object
is supposed to go. And this is what I have said in recent work (2009).7 However,
I have come to think that there are at least two better alternatives. The purpose
of this chapter is to explain what now seems to me problematic with the gappy
content proposal, to present the alternative views, and to draw out the conse-
quences of these reflections for the thesis of representationalism about phenom-
enal character.
3
Not everyone accepts this claim. One notable exception is Travis 2004. See also Brewer 2008.
4
On this view, the content is a structured entity. I should add that not all disjunctivists grant
that in cases of illusion, perceptual experiences have contents of the same sort as veridical perceptual
experiences. See, e.g., Martin 2006. Obviously, those disjunctivists who take this view cannot use the
present consideration to motivate their view.
5
McGinn (1982) and Davies (1992) deny that the content is singular for any perceptual
experiences. For criticisms of this position, see my 2009, ch. 4.
6
Assuming that the term ‘singular content’ is used in the usual way. For an opposing usage, see
Sainsbury 2006.
7
See also Bach (1997), Loar (2003), Burge (1991).
Imagistic and Possible-Word Content
294
Consider the singular content that object, O, is red. On the Russellian view, this con-
tent is complex, having object, O, and redness as its components.8 A visual experience
(or other mental state) having that content is accurate if and only if O is red. One way
to think of the Russellian content here—the content that O is red—is as an ordered
pair having O as its first member and redness as its second. Another way to think of
the content is as a structured, possible state of affairs built out of O and redness.
It was suggested above that where a visual experience is hallucinatory, the
content is just like a singular content but with a gap or a hole in it where an object
should go. But does this really make sense?9 On the ordered pair conception of
singular content, there must be two items to form the pair. Since a gap or a hole is
not an item, or so it seems, there is no first member of the ordered pair and so no
ordered pair at all. On the possible state of affairs conception, the relevant complex
is structured out of O and redness in the singular case. But in the gappy case, there
is no object O. So, how is there a complex entity structured out of its components?
A possible reply is to say that the missing item in both cases is the empty set.
Where one hallucinates, the content is a complex entity built out of the empty
set and various properties. But intuitively this is a bizarre proposal indeed. If the
empty set is the gap filler, then the hallucinatory experience is about the empty set.
So, in a hallucination, one experiences the empty set just as, in the veridical and
illusory cases, one sees object, O. Furthermore, if the hallucinatory experience is
about the empty set then it is experienced as being some way or other, for example,
red. So then the empty set looks red, just as the seen object, O, looks red when
one sees O and experiences it as red! This is more than a little hard to swallow.
Furthermore, the proposal effectively reconfigures hallucinations as illusions of a
special sort. Instead of there being no object perceived, there is now an object of
a special sort that is perceived, or if not perceived, at least experienced and this
object is not as it is experienced to be.
The empty set proposal is a desperate attempt to save the gappy content view.
It confuses the truth that in hallucination one does experience something—for
example, a ripe tomato—with the falsehood that there exists some thing one
experiences, the empty set being proposed as the relevant thing since no ordinary
object is available to do the job.
Another possible reply is that the missing item is an absence. This needs a
little explanation. Roberto Casati and Achille Varzi (1994) make an interesting
case for the view that holes exist. If, for example, I say truly that there is a hole in
the cheese, on their view, what makes my remark true is the existence of a hole
8
Obviously, there are other views of singular content. For ease of exposition, I shall stick with the
Russellian conception in what follows.
9
This question is a pressing one for gappy proposition theorists (e.g., Braun 2005) generally.
What is the Content of a Hallucinatory Experience? 295
surrounded by the relevant piece of cheese. What Casati and Varzi show (at a min-
imum) is that it is not at all easy to paraphrase away apparent commitment to holes
in our ordinary talk. Analogously, it might be suggested, a case can be made for the
view that there exist absences. Even if such a case were successful, however, there
is a pressing question for anyone who appeals to an absence in the context of a hal-
lucinatory experience, namely which absence is relevant?
For Keith, since he is hallucinating, there is no ripe tomato that he is seeing.
So, on the absence view, the relevant absence is that of any ripe tomato. That is
what he experiences as red. The content is an ordered pair of an absence of any ripe
tomato and redness, or a possible state of affairs structured out of these items,10
where the relevant absence presumably is a concrete entity existing wherever ripe
tomatoes do not and thus varying in its location with variations in the locations of
ripe tomatoes. But then how can Keith, sitting in his room, experience that entity?
How can that entity look any way to Keith?
Perhaps it will be replied that Keith experiences that entity by experiencing
part of it just as Keith sees a ripe tomato by seeing part of it (namely its facing
surface). If so, just which part of the absence does Keith experience? Again, at a
minimum, the proposal is very unpromising.
A third possible proposal is that the gap filler in the ordered pair is a
spatio-temporal region. On this proposal, in hallucinating a red object, one is
experiencing a particular spatial region as red. This seems plainly misguided,
however. Spatial regions do not look red, so one cannot experience them as red.
Perhaps it will be suggested that, in hallucination, one experiences a specific spa-
tial region as filled by an object having so-and-so properties, for example, redness.
The trouble now is that if one says this for the hallucinatory case, one should say
it for the veridical case too, at least if one wants to avoid the earlier mentioned
implausible and unlovely asymmetry. The price paid is that the seen object in the
veridical case is no longer in the content. In its place is the spatial region occupied
by the object. But that is not what one sees.
Another difficulty for the spatial region proposal is that it is very unclear
which region is to serve to fill the gap in the content. Suppose, for example, you
are in a dark room and you hallucinate a sudden bright pinpoint of light. It could
be that your experience does not locate the light at any particular distance away
from you. So, which spatial region in the room is the one that you are experiencing
as being occupied by the light? There is no obvious answer.
It seems to me that the net effect of these reflections is to cast serious doubt on
the wisdom of the idea that the gap in a gappy content should be filled. Whatever
else gappy content is for hallucinations, it can’t be a species of ordered pair content.
What are the remaining alternatives?
10
Obviously I am over-simplifying here, since the content of visual experience is extremely rich.
But this makes no difference for present purposes and so I ignore it.
Imagistic and Possible-Word Content
296
These remarks do not yet provide us with a clear account of what the gappy
content is supposed to be in the hallucinatory case. And here there remains a
puzzle. To appreciate this, note that in the open sentence ‘Fx’, there are two com-
ponent parts: the predicate ‘F’ and the variable ‘x’. The counterpart to ‘F’ in the
gappy content of a visual experience is the property of being F, as it is in the sin-
gular content case. What is the counterpart to the variable ‘x’? It looks as if we
have to say, as before, that it is something like a slot in a mailbox. Gappy content,
then, presumably is an ordered pair consisting of the empty slot and the relevant
property or complex of properties. But now we are back with a version of the or-
dered pair view and all of the obscurity that goes with it.11
Furthermore, if we do make this proposal about gappy content then, to pre-
serve uniformity, we ought to say that singular content is an ordered pair consisting
of the filled slot and the relevant property. Unfortunately, this is too general. Not
just any old filling of the slot will do in the singular case. We need one particular
filling—that provided by the seen object—so that the singular content is now more
complex than previously supposed, having something like the following structure
(where P is the relevant property and CONJ the truth function for conjunction):
< CONJ << the relation of filling, < a, the slot >>, < the property P, a >>>
Modifying the account of gappy content correspondingly, we have:
< AND << the relation of filling, < --, the slot>>, < the property P, -- >>>
where—is the absence of any seen object or some other dubious item. Obviously,
we are getting nowhere.
Perhaps we should think of the content of a hallucinatory experience as being
like a tree structure. The model in this case is the tree structure linguists take
sentences to have. On this proposal, in veridical and illusory cases, when one
experiences an object, O, as red, the structure has two branches, each of which
has an entity at the end:
Propositional Content
/ \
Object O being red
The left hand branch is reserved for objects and the right hand one for 1-place
properties. In the hallucinatory case, there are the same two branches but the left
hand one is empty:
Propositional Content
/ \
being red
11
Barwise and Perry (1981) call the propositional analogs of variables “indeterminates”. As applied
to the case of the content of hallucinatory experience, this does not help with the earlier worries.
Imagistic and Possible-Word Content
298
12
This oversimplifies King’s proposal minimally. For King’s motivations for such a view, see his
2007. King takes his account to explain the unity of a proposition as well as the constitutive link (as he
sees it) between propositions and language.
What is the Content of a Hallucinatory Experience? 299
in L encodes the instantiation function and b occurs at S’s right terminal node and
has the property of being red as its semantic value.
This does seem to me coherent. But it is complicated and it requires that
facts about language be part of the content of visual experience. This seems to me
counter-intuitive, especially if we accept (as I do) that experiences themselves are
more like maps or pictures than linguistic representations (see below). Perhaps
the proposal could be made more viable by allowing the relevant scheme of rep-
resentation to be internal and nonlinguistic in character; but obviously this would
necessitate further revisions since there would be no relevant sentential relation S.
The upshot, in my view, is that the prospects for understanding the putative
gappy content of visual experience are very gloomy. It behooves us to look for a
better alternative.
Visual experiences are not like conscious thoughts. One can think that there is a
ball on a box without one’s thought representing anything further about the ball
and the box; but one cannot have a visual experience of a ball on a box without
one’s experience representing such things as the color of the ball, the color of the
box, the relative size of the two, the shape of the box, the view-relative locations
of the ball and the box, and so on. Visual experiences, in representing one thing,
represent many. In this way, visual experiences are like maps or pictures whereas
thoughts are like sentences. Visual experiences are representationally rich. It is
tempting to infer from this that the content of visual experience must be corre-
spondingly rich. But this is a mistake. It falsely assumes that a property of the
vehicle of representation (the experience) must be a property of its content.
Structure in a representation need not be mirrored in structure in its content.
This point applies not just to experiences but to thoughts as well, even though
they lack the representational richness of experiences. Consider the thought that
Vulcan is a planet. The thought is complex, being composed of the concept Vulcan
and the concept planet, combined in a certain way (rather as the sentence “Vulcan
is a planet” is composed of the words ‘Vulcan” and ‘planet’ in a certain order).
However, the content of the thought need not be complex. The content can be just
a set of possible worlds—the set of worlds in which the referent of the concept
Vulcan has the property referred to by the concept planet (the property attributed
by the thought).13 Since the concept Vulcan has no referent either in the actual
world in any other possible world, the relevant set of possible worlds is empty.
Correspondingly, experiences are complex. They have representational parts.
Some parts represent objects seen (if there are any); others represent properties of
13
This is not to suggest that thoughts are to be individuated by their contents, so conceived. For
more on the nature of thought, see Sainsbury and Tye 2012.
Imagistic and Possible-Word Content
300
which the subjects of the experience are conscious. The latter representational parts
are arguably more like features (as is the case with real pictures), and so in one sense
they are not really parts at all but rather features of parts. But however this is devel-
oped further, a complexity in representational structure for experiences need not be
reflected in a corresponding complexity in representational content.
Once this point is appreciated, given the difficulties already encountered in
trying to understand gappy content, a natural suggestion is that the content of
a visual experience is simply a set of possible worlds, namely the set of worlds at
which the experience is accurate. On this view, the content of a visual experience
is unstructured in the sense that it has no component parts.
This suggestion preserves uniformity in content for all experiences. Experiences,
whether they are veridical, illusory, or hallucinatory, have associated with them an
appropriate set of possible worlds. An experience, thus, is accurate, if and only if the
actual world belongs to the appropriate set of possible worlds. Which is the appro-
priate set? Answer: the set of worlds at which the objects picked out by representa-
tional parts of the experience have the properties the experience aims to attribute to
those objects (however this is further cashed out). The objects thus picked out are
the objects (if any) that are seen. Where there are no seen objects, as in a hallucina-
tion, there are no possible worlds at which the objects picked out by the representa-
tional parts of the experience have the experienced properties. So, the set of worlds
associated with a hallucinatory experience is the empty set.14
What, then, of the reasons given at the beginning of the chapter for the view
that seen objects are components of the contents of visual experiences? Consider
first this reasoning from earlier on:
(1) If I see an object, it looks some way to me.
(2) If an object looks some way to me, then it is experienced as being some
way.
(3) If an object is experienced as being some way, then it is a component part of
the content of the experience (assuming that the experience has a content).
So,
(4) If I see an object, the seen object itself is a component part of the content
of the experience, assuming that experience is representational at all.
The premise I reject here is (3). If a given object, O, is experienced as being
some way then the experience represents O as being some way (e.g., red). But this
is now cashed out further in terms of the experience having as its representational
14
On this view, (rather obviously) experiences are not to be individuated simply by their
phenomenal character. The experience of seeing something, O, and experiencing it as red has a
different content (set of possible worlds associated with it) than does the experience of hallucinating
something red, even though the experiences have (or may have) the same phenomenal character. The
former experience has the set of possible worlds at which O is red as its content (and there are many
such worlds); the latter experience has the empty set as its content.
What is the Content of a Hallucinatory Experience? 301
content the set of possible worlds at which O is that way (red). And O is not part
of the content of the experience.
Of course, the set of possible worlds at which an experience that is about O
is accurate has members, namely possible worlds, each of which has O as a com-
ponent part (whether one supposes that worlds are maximal states of affairs or
(implausibly) concrete configurations of objects). But O is not a component part of
that set, any more than my heart is a component part of the set whose members are
me and my mother.15 So, O is not a part of the content of the experience.16
Even though, on my view, O is not a part of the content of the experience, the
experience remains crucially dependent on O. This is because the content of the
experience is specified by reference to O as the set of worlds at which O is red. Thus,
given that in actuality there is such an object as O, had things been different and O
not existed, neither would the experience occasioned in actuality by P’s seeing O.
The same points can be made in the case of thought on an austere view of
thought content. The thought that Cicero is an orator is a thought about Cicero.
It is about Cicero in virtue of its having an atomic nominative concept that refers
to Cicero. The thought attributes the property of being an orator to the person so
referred to. The content of the thought is the set of possible worlds at which Cicero
is an orator and the thought is true since the actual world is a member of that set.
That content does not contain Cicero as a component part even though the thought
is about Cicero. And that thought would not have existed had Cicero not existed.
The second piece of reasoning offered earlier for the view that the seen object
is in the content of the experience went as follows:
(5) In cases of illusion, the seen object appears other than it is.
(6) If the seen object appears other than it is, then the visual experience is
inaccurate.
The best explanation of such inaccuracy is
(7) Where there is a seen object, a visual experience has a content into which
the seen object enters along with its apparent properties so that the expe-
rience is accurate if and only if the object has those apparent properties.
I deny that (7) really is the best explanation of the relevant inaccuracy. If the
content of a given visual experience is a set of possible worlds at which the seen
object, O, has the properties, P1‒Pn, attributed to it by the experience then we have
15
Set membership should not be confused with the part-whole relation. The former is irreflexive,
asymmetric and intransitive; the latter reflexive, asymmetric and transitive.
16
Earlier, I assumed that the members of an ordered pair are parts of the ordered pair. This
assumption might be challenged on the grounds that ordered pairs can be defined set-theoretically
(e.g., on Wiener’s definition, <a,b>:= (((a), the empty set), ((b))), in which case they have sets as their
parts. The trouble here is that there are too many equally good definitions and thus too many equally
good candidates. We should accept all or none of them. We can’t accept all of them because they are in
conflict. So, we should accept none.
Imagistic and Possible-Word Content
302
an equally good explanation of the inaccuracy for the case in which O appears
other than it is. In that case, the actual world is not a member of the set of worlds
at which O has the properties, P1‒Pn and O is not a component part of the content.
I begin with some general remarks about indexicals and Kaplan’s theory of indexi-
cals (1989). Then I show its relevance to the case of visual experience. Indexicals
are terms that change their reference from utterance to utterance. Examples are ‘I’,
‘here’, ‘she’, ‘that’, ‘today’ and ‘here’.
Consider the following two utterances:
Tim: “I am hot”
Tom: “I am hot”.
Intuitively, these two utterances have the same linguistic meaning, but what Tim
says is different from what Tom says. Tim, who is cold (let us suppose) says some-
thing false; but Tom (who is hot) says something true. So, the content of Tim’s
remark is different from the content of Tom’s.
On Kaplan’s theory (1989), indexicals have contents with respect to contexts.
For example, the content of ‘I’ with respect to a given context C is the subject or
agent of C; the content of ‘that’ with respect to C is the object demonstrated in C;
the content of ‘here’ with respect to C is the location of C. The content of a sen-
tence containing an indexical is a structured proposition having as its constituents
the content of the indexical (the agent, place, object demonstrated, etc.) and the
contents of the other terms, where these contents are taken to be worldly enti-
ties: particulars, properties and relations. Thus, in the case of Tim’s utterance of
the sentence “I am hot,” the content of Tim’s remark is a structured proposition
containing Tim himself (the subject in this context) and the property of being hot
(the content of the predicate ‘is hot’17). The sentence is false in the context, given
that Tim is cold.
On Kaplan’s theory, the linguistic meaning of an indexical term is a function that
maps contexts onto contents, where the latter are those contents the term has at each
context. Kaplan calls this function the term’s character. Thus, consider the term ‘here’.
Its character is a function from contexts whose value at each context is the location
of that context. Similarly, the character of the term ‘that’ is a function from contexts
to the objects demonstrated in those contexts. In the case of sentences containing
indexicals, their characters are functions from contexts to the structured propositions
that are the contents of the sentences in those contexts.
17
We may say that it is the character of ‘is hot’ to refer to or express the property of being hot,
regardless of the context. The character here is a function that yields the property of being hot at every
context.
What is the Content of a Hallucinatory Experience? 303
18
For ease of exposition, I assume here a single object seen. Obviously, this is an over-simplification.
19
So, seen objects do indeed enter into the content of visual experience in such cases.
Imagistic and Possible-Word Content
304
12.4. Representationalism
20
Along with (most) other representationalists, I am happy to say that, in the hallucinatory
case, the perceiver is conscious of an un-instantiated property. This seems to me to be part of naïve
commonsense. Suppose that you had never seen any red things and then, one day, you hallucinated a
red car. Did you not then encounter redness in your experience? Did you not then “get a good look” at
redness (Hawthorne and Kovakovitch 2006), one that enabled you then and there to know what it is
What is the Content of a Hallucinatory Experience? 305
like to experience red? Of course, some further account needs to be given of what it is for an experience
to represent or be about a property that does not require that the property be instantiated. But there are
several such accounts in the literature. Take, for example, the Normal tracking account of basic sensory
representation (in first approximation, a sensory state is about a property, P, just in case the state is of
a type that is Normally tokened if and only if P is tokened and because P is tokened). This relationship
between the state and the property can obtain even if in some Abnormal situations the property is
not instantiated. The same is true if we think of property representation by experiences in terms of
indicator function (see Dretske 1995).
21
If experiences are like pictures, they have parts, features of which play a representational role
analogous to the role played by predicates in sentences; for present taxonomic purposes I am assuming
that this counts as a kind of predicative representation.
22
The thesis that visual experiences predicatively represent property complexes is also endorsed
by Mark Johnston (2004) and Colin McGinn (1999). Johnston writes of “sensible profiles” and McGinn
of “clusters of properties.” In my view, visual experiences are nonconceptual representations. (This is
another way in which visual experiences differ from thoughts.) So, where properties are conceptually
represented in perceptual acts, they are not represented by perceptual experiences proper but rather by
associated judgments or beliefs (within, for example, hybrid mental acts of seeing-that).
23
Likewise the visual experiences undergone in perceptually different waterfall illusion scenarios
do not present a problem for (R). Although the content of these experiences is the same—the empty
set—the property complex predicatively represented is different.
24
Given package 1, experiences that are exactly alike phenomenally can differ with respect to
their representational content (take a veridical experience and a phenomenally identical hallucinatory
one) and experiences that are exactly alike with respect to their representational content can differ
phenomenally (take phenomenally different hallucinatory experiences).
Imagistic and Possible-Word Content
306
Appendix
25
I ignore here more complicated cases in which there are multiple things seen. For some
discussion of these, see the appendix.
26
Of these two packages, the one I prefer is package 1. In part this is because of considerations of
systematic unity and fit with belief content (see here Sainsbury and Tye 2012).
27
Of course, these property complexes are much too coarse-grained to capture phenomenal
character anyway, since the experiences predicatively represent determinate shades of color,
determinate shapes, location and many other details. But this makes no difference to the point I am
currently making and so I ignore it.
What is the Content of a Hallucinatory Experience? 307
References
Bach, K. (1997). Searle against the world: How can experiences find their objects?
Manuscript. http://userwww.sfsu.edu/~kbach/Searle.html.
28
O1 and O2 jointly instantiate (P1). It is an interesting further question in metaphysics as to how
to understand what is involved in O1 and O2 having a property that is not possessed either by O1 alone
or by O2 alone or by the ordered pair <O1, O2>. Here are some further examples of joint instantiations:
John and Jane together lifted the piano; the children stood in a circle.
29
Thanks to Mark Sainsbury for written comments and Brian Cutter for discussion.
Imagistic and Possible-Word Content
308
Barwise, J., & Perry, J. (1981). Situations and Attitudes. Berlin: Walter de Gruyter.
Brewer, B. (2008). How to account for illusion. In A. Haddock & F. Macpherson (Eds.),
Disjunctivism: Perception, Action, Knowledge (pp. 168–180). Oxford: Oxford University
Press.
Burge, T. (1991). Vision and intentional content. In E. LePore and R. Van Gulick (Eds.), John
Searle and his Critics (pp. 195–214). Oxford: Blackwell.
Casati, R, & Varzi, A. C. (1994). Holes and Other Superficialities, Bradford Books. Cambridge,
MA: MIT Press.
Davies, M. (1992). Perceptual content and local supervenience. Proceedings of the Aristotelian
Society, 92, 21–45.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.
Hawthorne, J., & Kovakovitch, K. (2006). Disjunctivism. Proceedings of the Aristotelian
Society, Supplementary Volume, 80, 145–183.
Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 103, 113–183.
Kaplan, D. (1989). ‘Demonstratives’ and ‘Afterthoughts’. In J. Almog, J. Perry, & H. Wettstein
(Eds.), Themes from Kaplan (pp. 481–565). New York: Oxford University Press.
King, J. (2007). The Nature and Structure of Content. Oxford: Oxford University Press.
Loar, B. (2003). Phenomenal intentionality as the basis of mental content. In M. Hahn &
B. Ramberg (Eds.), Reflections and Replies: Essays on the Philosophy of Tyler Burge (pp.
229–258). Cambridge, MA: MIT Press.
Martin, M. G. F. (2006). On being alienated. In T. S. Gendler & J. Hawthorne (Eds.),
Perceptual Experience (pp. 354–410). Oxford: Oxford University Press.
McGinn, C. (1982). The Character of Mind. Oxford: Oxford University Press.
Price, H. H. (1932). Perception. London: Metheun & Co Ltd.
Sainsbury, M. (2006). Reference without Referents. Oxford: Oxford University Press.
Sainsbury, M., & Tye, M. (2012). Seven Puzzles of Thought (and How to Solve Them).
Oxford: Oxford University Press.
Tye, M. (2009). Consciousness Revisited: Materialism without Phenomenal Concepts.
Cambridge, MA: MIT Press.
PART FIVE
This chapter addresses each of two interlocking issues. If the two did not interlock,
the world would be a better place. But they do: What does vision represent? And,
the Representational theory of sensory qualities. I shall begin with the title question.
But, first: Does perception represent, at all? It seems to me and to many hard
to deny that it does. But that is because we tend to think mainly of vision, a very
one-sided diet of cases as Wittgenstein would have called it; that smell, taste and
touch represent is not at all obvious—though I have argued at some length that
they do (Lycan 1996, ch. 7). Moreover, there is a recent upsurge of Wittgensteinian
and/or Gibsonian and/or Naïve Realist and/or disjunctivist skepticism about
even vision’s being representational: Campbell (2002), Travis (2004), Noë (2005),
Brewer (2006), Fish (2009). I am unconvinced by those skeptics’ arguments,
but I cannot go into them here.1 I would agree with majority opinion that if any
sense modality represents, vision does. And I shall just assume that vision does
represent.
Of course, if vision and the other senses do not represent, the Representational
theory is a non-starter; and that would be very bad for us materialists.
In the spirit of traditional sense-datum theory, one might hold that vision repre-
sents only sense-datum-type properties, principally colors and shapes; compare
Marr’s “primal sketch.” Or one might allow that vision represents depth also, as
I am grateful to a number of people for help with this paper: to Nico Orlandi, Bill Fish, and
Susanna Siegel for (respective) conversations in 2006‒07 that inspired it; to Austen Clark, who gave
me very substantive help on it before commenting expertly on an early version presented to the
“Naturalized Philosophy of Mind and Language” conference in honor of Ruth Garrett Millikan,
University of Connecticut (October 2008); and to audiences at that conference and at the ANU, the
University of Otago, North Carolina State University, and the Tufts University Center for Cognitive
Science.
1
For effective critique, see Siegel (2010a) and Schellenberg (2011). 311
The Constituents of Perceptual Content and the Role of Perception
312
in Marr’s 2.5-D sketch, and volumetric shapes and distances as in Marr’s 3-D
sketch.
That is still a very spare inventory.2 Surely [but watch that “surely” operator!]
vision represents everyday objects, not just volumetric shapes and distances. And
object-recognition is obviously [!]one of vision’s functions. In The Modularity of
Mind (1983), Jerry Fodor defended the 3-D sketch view, but added that vision fur-
ther applies Rosch’s “basic categories,” such as “dog,” “shoe,” “chair,” “red,” “lady.”
The addition is incongruous with Fodor’s own strong modularity doctrine, which
requires pretty strict informational encapsulation. Perhaps he thinks, modularity
be damned, it is just common sense. But it is not Moorean common sense; it lies
somewhere between empirical psychology and the philosophy of representation.
Fodor does say that module outputs must be phenomenologically accessible, and
that the mere 3-D sketch does not satisfy that requirement. I am not sure why he
thinks the 3-D sketch is not phenomenologically accessible, but he defends the
requirement in a long footnote (1983, p. 136 n. 31).
One might add motion and change, as such.
There are further objections to such conservative views. (1) We simply see
and recognize individual things such as people (and their faces), not just prop-
erties or types of object. (Ruth Millikan has argued that “the ability to reidentify
things that are objectively the same when we encounter them in perception is
the most central cognitive ability that we possess” (2000, p. 109). (2) We sim-
ply see and recognize things as socially characterized—dollar bills, post offices,
square dances, without anything that could fairly be described as person-level
inference. Some liberals will go to the extreme, and maintain that we can
perceive (as such) electrons, social class, surges of monetary inflation, global
warming, and the like. (3) We perceive causal relations (Siegel 2006a, 2010b).
(4) Lyons (2005) defends a moderately and discriminatingly liberal view, based
on a highly original notion of “perceptual kinds” in the world. (5) There is evi-
dence that when we hear utterances or read text, we directly perceive meanings;
that is, our language modules are built to deliver sentence meanings and even
implicatures without any inference or other calculation from more primitive
percepts on the part of the whole subject (Pettit 2002). (6) Millikan argues for
an even broader field of direct perceivables: e.g., we can directly hear rain just
in virtue of someone saying, “It’s raining” (2004, p. 122); indeed, she says, “It’s
raining” is the sound of the rain from where we sit! (7) We see absences, as such
(Sorensen 2008; Farrenikova 2012).3
2
Clark (2000, 2004) suggests a conservative view, arguing that what vision primarily represents
are (a) features placed at locations and (b) “proto-objects” of the sort posited by Pylyshyn (2001). But
he does not rule out the notion that perceptual representation is layered (see below) and that vision
represents more sophisticated items at a higher level of organization.
3
Farrenikova points out that there is a difference between rich or nonconservative visual content
and conceptually high-level content. On her view, nonhuman animals can see absences.
What Does Vision Represent? 313
How might we adjudicate? That is, what should be our test for whether such-and-such
a thing or property is specifically represented in and by vision, without benefit of
either inference or some other contribution from general cognition?
We must be careful about this use of “vision.” Does it mean, (i) visual experience
as some writers put the question, (ii) seeing, whether “experienced” (i.e., consciously)
or not,4 or (iii) the visual system as investigated by cognitive and neuroscientists? I shall
try to mean (ii). (iii) would be an entirely empirical matter, though obviously scientific
results regarding the visual system bear fairly directly on our own issue. Moreover, I do
not doubt that the visual system subpersonally represents properties that are not seen
by the whole person whose visual system it is. (i) seems needlessly restrictive, since
(according to me) the difference between a type of mental state occurring noncon-
sciously and that same type of state occurring consciously is superficial, a matter of
whether the state is itself represented by a higher-order state (Lycan 1996, 1998, 2004);
it would not normally affect the state’s own representational content.
So my question more precisely is: what sorts of properties or things are spe-
cifically represented in and/or by person-level seeing? But at this point the issue
deepens horribly.
As Susanna Siegel has pointed out (2010b, ch. 2), negative arguments in this
area are ineffectual. But, I maintain, so are the most obvious positive arguments:
4
Some writers use “experience” liberally, to include all cases of seeing (or whatever sensing)
whether conscious or not. I prefer to reserve the term for sensing consciously, i.e., for sensings of which
their subjects are aware.
5
But neither could the Russellian riposte by appealing to the Argument from Illusion, even
though Russell thought the argument showed that belief in anything but sense-data required inference
The Constituents of Perceptual Content and the Role of Perception
314
from sense-data; the facts of illusion could equally be just misrepresentation by the visual system. For
some resistance to my phenomenological claim, recall Firth (1965).
6
Siegel (2006, sec. 2) goes into this matter more thoroughly, considering more than one way in
which psychosemantics might be thought to help answer our title question. But she argues convincingly
against each.
7
And it still lingers, at least among some older epistemologists: Hearing any apparent departure
from Naïve or at least very Direct Realism about perception, such a philosopher will reflexively accuse
the speaker of holding that we perceive sense-data, or representations, or retinal stimulation, etc.
8
Millikan’s discussion of directness suggests each of two senses. In one of them, “direct” means,
not mediated by inference, where “inference” is read fairly strongly as person-level conceptual activity.
Millikan believes there are many psychological processes in which intentional representations produce
further representations in a regular and indispensable computational way, but what is required for
actual inference is that the representations are the vehicles of the subject’s beliefs (2000, pp. 118–119).
Beliefs figure in socially recognized forms of inference, deduction, induction, abduction, not merely in
algorithmic internal processes; and beliefs are catholic and all-purpose, not dedicated. None has any
specific job to do, and none is isolated from other beliefs in any way. All that seems quite right to me,
as a good thing to mean by “noninferential,” and noninferential is one good thing to mean by “direct.”
In the other, much weaker sense of “direct” (2004, p. 162), it means, produced by a process
essentially involving no other intentional representations as opposed to merely natural signs in the
sense of Grice and Dretske. That too is a perfectly fine thing to mean by “direct,” though I do not think
Millikan went on to make much use of the notion.
What Does Vision Represent? 315
6. Siegel (2006a, 2006b, 2007, 2010b) has offered a new style of argument
for particular representational claims: the method of phenomenal contrast.
And, unlike the five preceding sorts of appeal, her style of argument
affords progress, though I shall argue that it too is ultimately inconclusive.
However, I will postpone discussion of Siegel’s method in order to address
a prior question.
rich technical senses those words have in our language, but neither am I speaking
purely de re as opposed to under some low-level common representation involv-
ing potential food.) Perhaps a bit less universal is a piece of meat. Then there is
gradually increasing cultural relativity. Likewise, (iii) even within a culture such
as that of the United States in 2012, abilities to see grow with background knowl-
edge, though not in the entirely fluid, freewheeling way alleged by Churchland.
Everyone sees food, as before; everyone sees clothing as such even if under no
very specific subclassification; ordinary landscape features; ordinary artifacts. But
a child may not see a tool as such, and even a mature person may not see a particu-
lar tool as the tool it is; at least some background knowledge is required. Likewise
for stars. And then come dollar bills, post offices, and square dances, and then
arcane chemical stains. . . .
If this gradualist view is correct, what of our question? It is not simply invali-
dated by false presupposition, as Churchland would have it. Rather, it is relativized
to processing stage, and is then empirical: Is property P represented in the small
core? If not, is P represented at stage Core+1? And so on. This approach takes the
issue out of the purview of introspection and of common sense, which, if my ear-
lier pessimistic discussion assessment was right, is a good thing.
I am going to assume that there are limits that stop short of complete
theory-ladenness and the most libertine uses of “looks.” I mentioned cars’
looking expensive and buildings’ looking pretentious. Maybe those are border-
line cases. But consider a house looking uninhabited; a person looking as if s/
he is suffering from Sartrean vertigo; a southern New Zealand peak looking as if
global warming has advanced; a bank manager looking as though the worldwide
recession has deepened; and local space looking Riemannian (to a physicist who
has truly internalized the General Theory of Relativity). I do not see how “vision,”
per se, could represent those. If we can broadly speaking see such things, it would
have to be because either (a) the freewheeling Kuhn et al. thesis is correct and our
question was misconceived, there being not even a relativized or gradualist dis-
tinction between “vision” and cognition, or (b) vision represents such things but
only derivatively, in virtue of representing simpler properties (see below). I shall
continue to reject the freewheeling thesis.
Siegel (2010b) appeals to the phenomenal change that occurs when vision is chun-
ked, when we substitute a more sophisticated recognitional capacity for what was
at first more painstaking. Her leading example is that of a novice forester who is
assigned a task to do with pine trees. At first the novice has to examine a tree to
see whether it is a pine tree, by comparing it to at least a short checklist of features.
But over time, the novice acquires a more direct recognitional ability and simply
spots pine trees as such. That is new phenomenology, or at least it contrasts with
What Does Vision Represent? 317
the original checklist phenomenology. It is tempting to say that the novice can
now see, hence visually represents, a kind, “pine tree,” and to conclude that vision
sometimes represents particular natural kinds.
If that were the argument, it would get nowhere, for familiar reasons. That
the novice can now pick out pine trees more readily and it feels different to her
to do so shows (without question-begging) nothing about her specifically visual
representata. A Marr-inspired conservative could grant this datum; for that
matter, so could a Russellian sense-datum theorist. But the foregoing is not Siegel’s
argument. Rather, Siegel (2006a, 2010b) proceeds by closing loopholes. Her master
argument is this (2010b, p. 101): Let E1 and E2 be, respectively, the before and after
types of visual experience exemplified by our novice’s original and better-trained
perceptions of pine trees, and let “K-properties” be a type of property denied by
the conservative—in our example, the property of being a pine tree. The overall
experience of which E1 is a part is the “contrasting” experience, and the overall
experience of which E2 is a part is the “target” experience.
(0) The target experience differs in its phenomenology from the contrasting
experience.
(1) If the target experience differs in its phenomenology from the contrast-
ing experience, then there is a phenomenological difference between E1
and E2.
(2) If there is a phenomenological difference between E1 and E2, then E1 and
E2 differ in content.
(3) If there is a difference in content between E1 and E2, it is a difference with
respect to K-properties represented in E1 and E2.
________________________
Anent (2), the competing explanations of the antecedent are changes that
are sensory but not representational. Siegel protests (pp. 109–110) that a “raw
feeling” occasioned by seeing the pine tree, such as one of familiarity, would
be extremely confusing were it not to represent something, a collection of
sense-data at least, as being familiar. (Siegel does not consider a “raw feel” of any
other type. I note that there are plenty of sensory feels or feel-components that
do not themselves represent, though according to me those must be identified
with functional properties; e.g., consider the motivational aspect of a pain expe-
rience (Lycan 1998).
Anent (3), the competing posit would be a visual representatum other than
a K-property. Siegel considers the example of a pine-tree-type shape Gestalt, the
idea being that what E2 represents that E1 does not is, not a tree, but a complex of
shapes that is generic enough to cover those of most pine trees but specific enough
to be shared by few non-pines. However, Siegel argues that although this Gestalt
move may work for pine trees, it does not extrapolate to other cases, such as a
human face’s property of expressing doubt (the faces of different sorts of people
may not express doubt in ways that can be captured in a shape Gestalt), or more
trenchantly, a line of Cyrillic text’s meaning.
At last we have an argument that does not beg the question and does a good
job of distributing the burden(s) of proof. And let us get rid of an apparent
objection: Siegel’s method does seem, short of question-begging, to frontload
liberalism, perhaps extreme liberalism. For we all have very high-level rec-
ognitional capacities that seem sensory and not to require inference. There is
obviously a phenomenological difference between seeing a car in economic
ignorance and having the very same car look expensive to you. And likewise for
ignorantly perceiving the Cyrillic text and seeing it with understanding, merely
seeing a face and seeing the same face as expressing doubt, etc. For that matter,
there is a phenomenological difference between practically anything and any-
thing. It should not be so easy to prove a priori that vision alone represents such
fancy properties.
Siegel (2007) addresses this, pointing out that the method of phenomenal con-
trast itself says nothing whatever about which properties are represented in vision;
the substantive work is done by premise (0), the phenomenology itself. It could
have been that Russell was firmly right, and there was no detectable phenomenal
difference between the novice’s experience of (what is in fact) the pine tree and
her/his learned experience of the tree as such; ditto for doubt in faces and mean-
ing in Cyrillic script. And pace premise (1), it could have been that there was no
sensory-phenomenal difference; etc. But the facts are otherwise. And the point of
Siegel’s master argument is to move us by discrete stages of reasoning from the
mere fact of phenomenal difference to the conclusion that a K-property is visually
represented.
Yet there are problems.
What Does Vision Represent? 319
By a “layering” view I mean one according to which one and the same representa-
tion has multiple (at least double) contents, systematically related to each other by
some asymmetrical priority relation.
9
Since writing the previous draft of this paper, I have learned from Wesley Sauret that monitoring
and attention are not as similar as they sound, and that majority opinion in neuroscience is that attention
is not in itself intentional. (More on this in a joint paper, in preparation.) It remains true that attention
alters the first-order state and the relevant sensory qualities; see Block (2010) and Lycan (forthcoming).
10
For further criticism of Siegel’s argument, see Lyons (2005).
The Constituents of Perceptual Content and the Role of Perception
320
11
Arguments for that crucial claim: (1) Any item is a representation only if it is used by some
cognizer as one; a representation represents something to someone or something. Since many of our
internal representations are subpersonal and do not represent anything to us ourselves, they must
represent to our subpersonal agencies. (2) On general grounds, Millikan is a teleofunctionalist and
she also—famously—gives a theory of teleology in terms of selection for effects; so representations as
such must be characterized in terms of their characteristic effects, and effects are relative to consuming
subsystems. (3) The consumer theory gives the best solution to the well-known indeterminacy problem,
most often exemplified by the question of whether the frog’s distinctive tongue-zap-stimulating
representation has black dots, small black things, flies, frog-food, edibles generally, or something else,
as its intentional object. (4) The theory handles all other data better than do competing views.
12
But I do not think Millikan herself believes in layering. She does maintain that higher-level,
more abstract representations are systematically and automatically produced by lower-level ones, a
process she terms “translation” and contrasts (as before) with inference. But the translation relation
holds between distinct, indeed successive representations: You represent some edges and shadings,
and (causally) respond to that by then representing a face-shape, and respond in turn to that in the
context by representing Johnny (2000, ch. 4 and p. 113). She does not use any such expression as “by”
or “in virtue of.”
She does have the excellent notion of “levels of distality”:
Contrary to much of the philosophical tradition, there is no single level of the outer world, such
as physical objects versus the mere surfaces of physical objects, or such as the presence of certain
phyical [sic] objects or of events versus mere sounds, of which the eyes or the ears are designed
exclusively to produce direct representations. Depending on the animal’s needs, various levels
of distality of direct perception may be mediated by the same sensory end organs. The affairs
naturally signified by retinal patterns, vibrating ear drums, stimulated odor sensors, and so forth,
are at various distances and are mediated in diverse ways. (2004, p. 162)
But that does not entail layering.
What Does Vision Represent? 321
Schellenberg’s view is superior to mine, in that it does not require any element of
illusion or seeing things that are not really there. In fact, I recant my 1996 view in
favor of hers.
(Of course all these views continue simply to assume that vision does represent.)
I have characterized layering theses by reference to a single representation’s
having multiple contents. There is a slightly less ambitious thesis available: that
an original representation having conservative or low-level content immediately
causes a second that has next-level, more sophisticated content, and the second
immediately causes a third, . . . , all too rapidly for introspection to detect that
the layered representations are numerically distinct. Some may prefer this ver-
sion. I continue to prefer the original, though I am not sure how we might decide
between the two.
Layering views offer some compromise with Siegel and other liberals. It seems
likely that although K-properties are not represented in the lowest layer, they may
be represented by something else’s having been represented in that layer or an
intermediate one.
But things get worse again. Things, because the original question of what
vision represents is about to be aggravated, and there are now more troublesome
apparent counterexamples to the Representational theory. I turn to matters of
aspect-perception.
13.7. Aspect-Perception
Different subspecies may need different treatments. The obvious big difference is
between seeing an object as F and seeing a picture as a picture of something F.
The late Richard Wollheim (1996) made a valuable distinction between seeing-as
and “seeing in.” It is less that we see the duck-rabbit figure as a picture of a duck
(though of course we can do that) than that we see a duck and alternately a rabbit
in the picture. And it is fairly clear that there is a stricter sense in which (unless
visual conditions are very unusual) an ordinary wineglass cannot be seen as a ship
of the line.
Wittgenstein’s mystery was this: Seeing an aspect seems a voluntary matter
of interpretation; yet it is not merely interpretation, a cognitive construction
put on what is strictly seen, but itself a kind of seeing, a specifically visual
phenomenon that (as Wittgenstein would not have put it) is part of visual
phenomenology.
So—again—what does vision represent? Does it represent a coconut when
Penrod sees Georgie Bassett’s head as one? (In the story, the drowsy Penrod did
sincerely take Georgie’s head to be a coconut, but we should also consider the case
where Penrod is aspect-flipping at will and whimsically chooses to see it as one.)
I am reasonably sure that in the second case at least, Penrod’s visual system does
not represent the property of being a coconut, and I am even more sure that in
the Aubrey case the officers’ visual systems do not represent wineglasses as being
ships, but I could not confidently defend those views against the rhetorical ques-
tion: If seeing-as is a specifically visual and not merely cognitive phenomenon,
then how is seeing a head as a coconut not, at least in part, visually representing
the head as a coconut?
“Sensory qualities” as I shall use the term are the distinctive introspectible qualita-
tive properties given in sensory experience—colors, smells, tastes, textures. A par-
adigm would be the color of an after-image, or that of a hallucinated object, to
emphasize that the qualities I am talking about are subjective, not or not necessarily
real features of real objects in the external world.
What Does Vision Represent? 323
13
Two side issues: First, Macpherson’s official target is only a Representational theory, Tye’s
(1995), that restricts itself to nonconceptual content. She simply ignores layering views, not to mention
a Representationalism based on Fodor/Rosch.—But then, we do not know that vision does represent
ordinary objects.
Second: In the matter of psychosemantics, Macpherson again singles out Tye and (in passing)
uses his particular covariation psychosemantics against him. That is his problem, not one for
Representationalism generally.—But then, if there is any correct psychosemantics for vision, it is
probably comparably simple.
14
There mellifluously called a “Q-property.”
The Constituents of Perceptual Content and the Role of Perception
324
A B
does not seem to be that of feature detection, and if not, then I would argue that
aspect examples could not refute the Representational theory of sensory qualities.
But I shall ignore this and assess Macpherson’s cases on her own terms. Here is her
opening display.
Peacocke claims that his layering view handles the Mach square/diamond,
and without resorting to higher-level properties such as “diamond.” He appeals to
representation of the “symmetry-about” relation: when we see a square as such, vi-
sion represents the figure as symmetrical about the bisectors of its sides, but when
we see the same figure as a diamond, vision represents a different property, that of
symmetry about the bisectors of its angles. Macpherson makes several objections,
of which the most trenchant is that some ambiguous figures are not symmetrical
at all:
Up Up
Right
Left Right
Left
Down
Down
FIGURE 13.3
Thus, the visual experience may represent properties that parts of the figure
have in relation to the axes, such as “having an angle pointing directly up.”
Macpherson objects that this fails to account for the shape constancy of un-
ambiguous figures. Representationalists “must explain either why only one set
of axes can be imposed upon certain figures (the non-ambiguous ones) and why
more than one set can be imposed upon other figures (the ambiguous ones), or
they must explain why the experiences that represent different properties do not
give rise to experiences with different phenomenal characters in non-ambiguous
figures, but do so in the other ambiguous figure cases” (p. 108).
The former at least does need explaining. But I do not agree that the burden is
on the Representationalist. What first and foremost needs explaining is why some
figures are ambiguous and others not, and that is everyone’s problem. Once that
explanation is provided, the Representationalist can (probably) buy into it.
In any case I am inclined to take seeing-as to be an attentional phenomenon
(Chastain and Burnham 1975; Ricci and Blundo 1990; Kleinschmidt et al. 1998).
That at least reduces Wittgenstein’s mystery to a slightly more tractable puzzle about
attending: that visual attending in particular is a visual phenomenon, in that at least
it affects visual phenomenology; but in some sense what is seen does not change.
As noted above, Christopher Peacocke, Bernhard Nickel and Ned Block have
offered cases involving shifts of attention as counterexamples to the Representational
theory. I have addressed those elsewhere (Lycan 2000, forthcoming), and will not
repeat those discussions here. I offered a number of options for accommodating
the examples within the Representationalist framework, and I now suggest that
most of those options will apply back to cases of aspect perception.
But what of representational content in aspect perception? The obvious sug-
gestion is that when Penrod sees Georgie’s head as a coconut, his visual state
represents a coconut. Certainly somewhere in the experience—especially in
Tarkington’s original version as opposed to the voluntary-flipping version—the
The Constituents of Perceptual Content and the Role of Perception
326
13.10. Conclusions
I can draw hardly any. My main thesis remains, that it is hard to see what could
establish a claim as to what vision does or does not represent. Another thing that
remains is the possibility that the question is a bad one. All this, I fear, is grist to
the mill of the skeptics who deny that vision represents in the first place. I probably
should not have written this chapter.
References
Block, N. J. (1996). Mental paint and mental latex. In E. Villanueva (Ed.), Philosophical
Issues, 7: Perception (pp. 19–40). Atascadero, CA: Ridgeview Publishing.
Block, N. J. (2003). Mental paint. In M. Hahn & B. Ramberg (Eds.), Reflections and
Replies: Essays on the Philosophy of Tyler Burge (pp. 165–200). Cambridge, MA: MIT
Press.
Block, N. (2010). Attention and mental paint. In E. Sosa & E. Villanueva (Eds.), Philosophical
Issues, 20: Philosophy of Mind (pp. 23–63). Oxford: Wiley-Blackwell.
Brewer, B (2006). Perception and content. European Journal of Philosophy, 14, 165–181.
Campbell, J. (2002). Reference and Consciousness. Oxford: Oxford University Press.
Chastain, G., & Burnham, C. A. (1975). The first glimpse determines the perception of an
ambiguous figure. Perception and Psychophysics, 17, 221–224.
Churchland, P. (1988). Perceptual plasticity and theoretical neutrality: A reply to Jerry
Fodor. Philosophy of Science, 55, 167–187.
Clark, A. (2000). A Theory of Sentience. Oxford: Oxford University Press.
Clark, A. (2004). Feature-placing and proto-objects. Philosophical Psychology, 17, 451–477.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: Bradford Books / MIT Press.
Farrenikova, A. (2012). Seeing absence. Philosophical Studies, online, November.
Ferrante, D., Gerbino, I., & Rock, I. (1997). The right angle. In I. Rock (Ed.), Indirect
Perception (pp. 163–70). Cambridge, MA: MIT Press.
What Does Vision Represent? 327
Firth, R. (1965). Sense-data and the percept theory. In R. J. Swartz (Ed.), Perceiving, Sensing,
and Knowing (pp. 204–70). Garden City, NY: Anchor Books.
Fish, W. (2009). Perception, Hallucination, and Illusion. Oxford: Oxford University Press.
Fodor, J. A. (1983). The Modularity of Mind. Cambridge, MA: Bradford Books / MIT Press.
Gilman, D. (1992). A new perspective on pictorial representation. Australasian Journal of
Philosophy, 70, 174–186.
Gilman, D. (1994). Pictures in cognition. Erkenntnis, 41, 87–102.
Kleinschmidt, A., Buchel, C., Zeki, S., Frackowiak, R. S. (1998). Human brain activity
during spontaneously reversing perception of ambiguous figures. Proceedings of the
Royal Society of London, Series B: Biological Sciences, 265, 2427–2433.
Lycan, W. G. (1996). Consciousness and Experience. Cambridge, MA: Bradford Books / MIT
Press.
Lycan, W. G. (1998). In defense of the representational theory of qualia (replies to Neander,
Rey and Tye). In J. Tomberlin (Ed.), Philosophical Perspectives, Vol. 12: Language, Mind
and Ontology (pp. 479–87). Atascadero, CA: Ridgeview Publishing.
Lycan, W. G. (2000). Representational theories of consciousness. In E. N. Zalta (Ed.),
The Stanford Encyclopedia of Philosophy. Summer 2000 edn.; extensively revised
and greatly expanded editions, 2004, 2006, 2012. http://plato.stanford.edu/entries/
consciousness-representational/.
Lycan, W. G. (2004). The superiority of HOP to HOT. In R. Gennaro (Ed.), Higher-Order
Theories of Consciousness (pp. 93–113). Amsterdam and Philadelphia: John Benjamins.
Lycan, W. G. (2008). Phenomenal intentionalities. American Philosophical Quarterly, 45,
233–252.
Lycan, W. G. (forthcoming). Block and the representation theory of sensory qualities. In A
Festschrift for Ned Block, ed. A. Pautz & D. Stoljar. Cambridge, MA: MIT Press.
Lyons, J. (2005). Clades, Capgras, and perceptual kinds. Philosophical Topics, 33, 185–206.
MacPherson, F. (2006). Ambiguous figures and the content of experience. Noûs, 40, 82–117.
Millikan, R. G. (1989). Biosemantics. Journal of Philosophy, 86, 281–297.
Millikan, R. G. (2000). On Clear and Confused Ideas. Cambridge: Cambridge University
Press.
Millikan, R. G. (2004). Varieties of Meaning. Cambridge, MA: Bradford Books / MIT Press.
Nickel, B. (2007). Against intentionalism. Philosophical Studies, 136, 279–304.
Noë, A. (2004). Action in Perception. Cambridge, MA: MIT Press.
Noë, A. (2005). Real presence. Philosophical Topics, 33, 235–264.
Orlandi, N. (2011a). Ambiguous figures and representationalism. Phenomenology and the
Cognitive Sciences, 10, 307–323.
Orlandi, N. (2011b). The innocent eye: Seeing-as without concepts. American Philosophical
Quarterly, 48, 17–32.
Orlandi, N. (forthcoming). Embedded seeing-as: Multi-stable visual perception without in-
terpretation. Philosophical Psychology.
Peacocke, C. (1983). Sense and Content. Oxford: Oxford University Press.
Peacocke, C. (1992). A Study of Concepts. Cambridge, MA: MIT Press.
Pettit, D. (2002). Why knowledge is unnecessary for understanding language. Mind, 111,
519–550.
Pylyshyn, Z. (2001). Visual indexes, preconceptual objects, and situated vision. Cognition,
80, 127–158.
The Constituents of Perceptual Content and the Role of Perception
328
Ricci, C., & Blundo, C. (1990). Perception of ambiguous figures after focal brain lesions.
Neuropsychologia, 28, 1163–1173.
Schellenberg, S. (2008). The situation-dependency of perception. Journal of Philosophy, 105,
55–85.
Schellenberg, S. (2011). Perceptual content defended. Noûs, 45, 1–37.
Segall, M. H., Campbell, D. T., & Herskovit, M. J. (1968). The influence of culture on visual
perception. In H. Toch and C. Smith (Eds.), Social Perception (pp. 139–44).
Siegel, S. (2006a). Which properties are represented in perception? In T. Gendler & J.
Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford University
Press.
Siegel, S. (2006b). Are Kind Properties Represented in Perception? Manuscript.
Siegel, S. (2007). How can we discover the contents of experience? Southern Journal of
Philosophy, 45, 127–142.
Siegel, S. (2010a). Do visual experiences have contents? In B. Nanay (Ed.), Perceiving the
World (pp. 333–68). Oxford: Oxford University Press.
Siegel, S. (2010b). The Contents of Visual Experience. Oxford: Oxford University Press.
Sorensen, R. (2008). Seeing Dark Things: The Philosophy of Shadows. Oxford: Oxford
University Press.
Travis, C. (2004). The silence of the senses. Mind, 113, 57–94.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: Bradford Books/MIT Press.
Wittgenstein, L. (1953). Philosophical Investigations. Oxford: Basil Blackwell.
Wollheim, R. (1996). Seeing-as, seeing-in, and pictorial representation. In Art and its
Objects (pp. 205–25). Cambridge: Cambridge University Press.
14
1
Both were posed by an anonymous referee, who asked that I address them.
Phenomenal Intentionality and Secondary Qualities 331
become a “live” one? I think that one important factor is the following. Until late
in the twentieth century, a very dominant doctrine in the philosophy of mind
was what Graham and Tienson and I call “separatism”—the idea that the phe-
nomenal aspects of conscious mentality are non-intentional, whereas the inten-
tional aspects of conscious mentality lack proprietary phenomenal character.
(Here, ‘conscious’ essentially means not unconscious; fans of separatism typically
embrace Ned Block’s influential distinction (Block 1995) between being phenom-
enally conscious and being “access conscious.”) On an orthodox version of sepa-
ratism, sensory-perceptual experiences were regarded as lacking representational
content (despite having so-called “phenomenal character”), whereas the percep-
tual beliefs that arise from sensory-perceptual experience were regarded as lacking
proprietary phenomenal character. In recent years, however, separatism has been
called into question by two influential developments. One is the rise of so-called
“representationalism,” which (i) treats phenomenal character (including the phe-
nomenal character of sensory-perceptual experience) as being intentional, and (ii)
embraces one or another externalist, reductive, account of mental intentionality
itself.2 The other is the rise in popularity of what Uriah Kriegel (2013) has dubbed
“the phenomenal intentionality research program”; those within this camp, myself
included, contend that the phenomenal character of sensory-perceptual experi-
ence is inherently intentional, and that this intentionality is narrow and intrinsic
(rather than conforming to any externalist account).3
Second, what is at stake in this debate? In my view, quite a lot—but I will
briefly mention just a few issues. First is the nature of sensory-perceptual phe-
nomenal character itself. (I claim that it is intentional, and intrinsically so.)
Second is whether phenomenal character in general, and sensory-perceptual
phenomenal character in particular, can be “naturalized” via some kind of
externalist, reductive, psycho-semantics. (I claim that it cannot be naturalized
that way at least, and perhaps cannot be naturalized at all—although I remain
a “wannabe materialist.”) Third is whether, and if so how, sensory-perceptual
experience provides evidential justification for the perceptual beliefs to which
such experience gives rise. (I claim that typically it does, in part because
perceptual beliefs inherit their content fairly directly from the content of
sensory-perceptual phenomenal character.4)
2
See, for instance, Dretske (1995), Tye (1995), and Lycan (1995). In my view, when fans of this
approach appropriated the label ‘representationalism’, they were engaging (perhaps unwittingly) in
terminological tyranny.
3
This view is no less deserving of the label ‘representationalism’ than is so-called representationalism.
That’s why those who have appropriated this label are guilty of terminological tyranny.
4
Jack Lyons (2009) maintains that sensory-perceptual phenomenal character is either entirely
non-intentional or at best extremely thin in its intentional content, and hence that sensory-perceptual
experience cannot provide evidential justification for perceptual beliefs. He defends a non-evidentialist,
reliabilist account of the justificatory status of perceptual beliefs. In Horgan (2011) I argue that Lyon’s
approach presupposes an objectionable form of separatism.
The Constituents of Perceptual Content and the Role of Perception
332
conditions. But whereas the narrow truth conditions of your own beliefs are largely
satisfied, those of your BIV physical duplicate’s matching beliefs largely fail to be
satisfied; thus, the BIV’s belief system is systematically nonveridical.
On the other hand, an exact match in narrow content between your own
intentional mental states and the corresponding states in your BIV physical
duplicate does not require or involve an exact match in referents (if any) of all the
various matching, putatively referring, thought-constituents. For instance, some
of your own occurrent thoughts that you would express linguistically using par-
ticular proper names—say, the thought that Michele Bachmann is not a genius—
involve singular thought-constituents whose referents (if any) are determined
partly in virtue of certain external relations that obtain between you and those
referents. Thus, your occurrent thought that Michele Bachmann is not a genius
involves a singular thought-constituent that purports to refer to a particular spe-
cific person (viz., Michele Bachmann); its actually referring, and its referring to
the specific individual to whom it does refer, depends upon there being certain
suitable external relations linking you to a unique eligible referent (viz., Michele
Bachmann). A Twin-Earthly physical duplicate of yourself, in a Twin-Earthly
duplicate local environment, would refer to a different individual (viz., Twin-
Michele) via the corresponding singular thought-constituent of the corresponding
occurrent thought. And, in the case of your BIV physical duplicate, the matching
singular thought-constituent fails to refer at all, because the BIV does not bear
suitable externalistic relations to any suitably reference-eligible individual in its
own actual environment. (Parallel remarks apply to thought-constituents that pur-
port to refer to natural kinds, such as the thought-constituent that you yourself
would express linguistically with the word ‘water’.)
For mental states involving thought-constituents for which reference depends
upon externalistic factors, there are two kinds of intentionality, each involving its
own truth conditions. First is the kind of intentionality already mentioned above:
phenomenal intentionality, with truth conditions that are phenomenally consti-
tuted and narrow. Second is externalistic intentionality, with wide truth conditions
that incorporate the actual referents (if any) of the relevant thought-constituents.
Your own thought that Michele Bachmann is not a genius, and the corresponding
thoughts of your BIV physical duplicate and your Twin Earth physical duplicate,
have matching phenomenal intentionality, with matching truth conditions. (These
truth conditions are satisfied in your case and in the case of your Twin Earth du-
plicate, but not in the case of your BIV duplicate.) On the other hand, your own
thought that Michele Bachmann is not a genius and your Twin Earth duplicate’s
corresponding thought do not have matching externalistic intentionality, because
the externalistic truth conditions of these respective thoughts do not match: the
truth value of your own thought depends upon the intelligence level of Michele
Bachmann, whereas the truth value of your Twin Earth duplicate’s corresponding
thought depends upon the intelligence level of an entirely different individual, viz.,
Twin-Michele. (Each thought’s wide truth conditions are indeed satisfied.) As for
The Constituents of Perceptual Content and the Role of Perception
334
your BIV duplicate’s thought, it lacks externalistic intentionality and wide truth
conditions, because its singular thought-constituent which purports to refer to a
person called ‘Michele Bachmann’ does not actually refer at all.
Our account rests heavily and essentially upon two key contentions.
First, mental reference to many properties and relations—including var-
ious spatiotemporal-location properties, shape-properties, size-properties,
artifact-properties, and personhood-involving properties—is wholly constituted
by phenomenology alone. Even systematically nonveridical phenomenology, as in
the case of the BIV, provides reference-constituting experiential acquaintance with
such properties and relations. It makes no difference to such experiential acquaint-
ance with such properties—and hence it makes no difference to mental reference
to such properties—whether or not the properties with which one becomes expe-
rientially acquainted are ever actually instantiated in one’s ambient environment.
Second, in the case of thought constituents whose reference (if any) depends
constitutively upon certain externalistic elements, the mechanisms of reference-
fixation crucially involve phenomenally constituted grounding presuppositions
(as we call them). Thus, phenomenal intentionality is more basic than exter-
nalistic intentionality, since the latter depends in part upon the former (as well
as depending, in part, upon externalistic factors). Suppose, for example, that
you have an occurrent thought that you could express linguistically by saying
“That picture is hanging crooked,” where the singular thought-constituent
expressible linguistically by ‘that picture’ purports to refer to a picture on the
wall directly in front of you. This thought-content involves phenomenally consti-
tuted grounding presuppositions that must be satisfied in order for the singular
thought-constituent to refer: roughly, there must be an object at a certain location
relative to yourself (a location that you could designate linguistically by a specific
use of the place-indexical ‘there’), this object must be a picture, there must not be
any other picture at that location that is an equally eligible potential referent of
‘that picture’, and this object must be causing your current picture-experience. If
these grounding presuppositions are satisfied by some specific concrete particu-
lar in your ambient environment—some particular object that is a picture and is
uniquely suitably located—then your singular thought-constituent thereby refers
to that very object. Which object your thought-constituent refers to, if any, thus
depends jointly upon two factors, one phenomenally constituted and one external-
istic: on the one hand, the phenomenally constituted grounding presuppositions,
and on the other hand, the unique actual object in your ambient environment that
satisfies those presuppositions.
The GTH approach, as described in the preceding section, leaves open various
questions about phenomenal color-intentionality and about the metaphysics of
Phenomenal Intentionality and Secondary Qualities 335
color. Are color-properties among those with which one is immediately acquainted
in experience in such a way that one can forthwith refer to these properties men-
tally even if one is an envatted brain? Or is it rather the case that mental reference
to color-properties incorporates a constitutive externalistic aspect, so that one’s
BIV’s thoughts purporting to refer to color-properties fail to refer at all? Are color-
properties ever actually instantiated in the world? If they are instantiated, does
color-experience represent their real nature?
Before taking up such questions, let me introduce a distinction that will prove
useful below, between two kinds of intentional content. On one hand is presenta-
tional content; it accrues directly to perceptual experience itself, whether or not
the experiencing subject forms a perceptual belief corresponding to that content.
(For example, when one is presented with an instance of the Müller-Lyer illusion
in a situation where one firmly believes that the two horizontal lines are the same
length, one of those lines nonetheless still looks longer than the other; thus, one’s
perceptual presentational content is as-of one horizontal line being longer than
the other, even though one does not form the corresponding perceptual belief.)
On the other hand is judgmental content: it accrues to full-fledged judgments,
including those that arise directly from perceptual experience.5
The distinction between presentational intentional content and judgmental
content is orthogonal to the distinction between phenomenal intentionality and
externalistic intentionality. Perceptual experience can have both phenomenal
presentational content (which is fully constituted by perceptual-experiential phe-
nomenology alone) and externalistic presentational content (which is constituted
in part by perceptual-experiential phenomenology and in part by suitable exter-
nalistic connections). An example of the latter would be a visual-presentational
experience as-of Sarah Palin—an experience which could occur even if one lacks
the corresponding belief. (Perhaps one is watching Saturday Night Live on televi-
sion, and one believes that one is really seeing Tina Fey.) Likewise, mental states
with judgmental content—e.g., occurrent beliefs—can have both phenomenal
judgmental content and externalistic judgmental content.
With the distinction between presentational and judgmental content at
hand, one potential way of extending the GTH approach to address phenomenal
color-intentionality and the metaphysics of color would be to embrace the fol-
lowing claims.
(1)
Phenomenology directly acquaints the experiencing subject with
color-properties, in such a way that the presentational phenomenal
5
For more on the distinction between presentational content and judgmental content, see
Horgan (2007). I take this distinction to be orthogonal to the distinction between conceptual and
non-conceptual content. I understand the latter distinction in terms of whether or not the given aspect
of presentational content is something for which one has concepts. Judgmental content is conceptual;
but presentational content can be conceptual, or nonconceptual, or a mixture of both. (And judgmental
content can include indexical conceptual elements).
The Constituents of Perceptual Content and the Role of Perception
336
6
Talk of what color experiences are like is derivative: color experiences present colors as
something-they-are-like properties—i.e., sensuous or phenomenal properties—of external objects. The
experientially manifest nature of these properties constitutes what they are like (as given in experience).
See Maund ([1997] 2006)—especially section 3, “The Natural Concept of Color.” What Maund calls the
“natural concept” of color picks out what, on my proposal, constitutes presentational color-properties.
Phenomenal Intentionality and Secondary Qualities 337
queer-looking from the scientific perspective—a fact that was clear right at the
dawn of modern science, and has remained clear ever since. Science has no the-
oretical need for them and no natural way of incorporating them. The scientific
worldview has brought about a “fall from Eden” (as David Chalmers puts it, in
Chalmers 2006) as far as experientially presented color-properties are concerned.7
7
As best I can tell from the philosophical and scientific literature about color that I am familiar
with, the extant position concerning colors and color-content that is closest to the one I will advocate
below is the position set forth in Chalmers (2006). Commenting on similarities and differences
between that position and mine is a task I leave for another occasion.
The Constituents of Perceptual Content and the Role of Perception
338
8
Since the proposal concerns phenomenal intentionality, it applies to a brain-in-vat duplicate
of an ordinary human experiencer in the same way it applies to the human experiencer. More on this
below, in section 14.6.
Phenomenal Intentionality and Secondary Qualities 339
normally are very well warranted by the visual-perceptual experiences that gen-
erate them, viz., visual-perceptual experiences as-of the instantiations of the cor-
responding, presentational, color-properties. And yet this account also honors the
introspectively evident phenomenology of visual-perceptual experience by recog-
nizing that such experience presents color-properties not as dispositional at all,
but rather as intrinsic, objective, features of the surfaces of external objects.
Admittedly, this account of color-content is an error theory of sorts: it treats
presentational color-content as systematically non-veridical. Visual experience
presents us with an apparently Edenic world, an apparent world in which pres-
entational color-properties are indeed instantiated on the surfaces of external
objects—and there is excellent scientific reason to believe that our world is not ac-
tually Edenic at all. But there is no serious theoretical cost in embracing this kind
of error theory, because doing so just amounts to embracing that scientific moral
while still acknowledging the actual, introspectively obvious, phenomenological
character of visual color-experience. It would be much more theoretically costly
to embrace an error theory asserting that everyday color-attributing judgments,
and the statements that express such judgments, are systematically and radically
false. But that high cost is precisely what my proposal avoids via the device of dis-
tinguishing between presentational color-content and judgmental color-content.
Admittedly too, this account goes contrary to the idea, which initially seems
very plausible, that the content inherited by color-judgments from the corresponding
color-experiences is literally identical to the presentational color-content of the
experiences themselves. But that idea, although quite natural from the perspective
of naïve common sense, has an epistemic status that is deeply hostage to how things
go in empirical science. Naïve common sense also cleaves very strongly—more
strongly, really—to the idea that visual color experience normally provides over-
whelmingly strong epistemic justification for everyday color-attributing judgments.
Given the pertinent science, something has to give, as far as naïve common sense
is concerned. I submit that on balance, in terms of comparative theoretical costs
and benefits, what should give is the naïve common-sensical belief that presenta-
tional color-content and judgmental color-content are identical. Once that belief is
jettisoned, everyday color-judgments turn out to be frequently true after all: objects
really do possess judgmental color-properties, and these are the properties that are
attributed to them by color-judgments.
It bears emphasis why, and how, the availability of the account of colors and
color-content here proposed depends upon the wider GTH framework. For some
time now in philosophy of mind, the dominant approaches to mental intentionality
have been one or another version of a view which Graham and Tienson and I call
strong externalism: roughly and generically, the view that all mental intentionality
Phenomenal Intentionality and Secondary Qualities 341
neither the inverts nor anybody else are even aware that there are inverts among
the human population. (The differences are detectable in principle, via the right
kinds of third-person monitoring of neural activity in the visual cortex of an invert
while presenting the invert with objects of various colors; but so far no such moni-
toring has been carried out.) Given my account of presentational color-content
and judgmental color-content, what should one say about the inverts, concerning
the content and the truth values of their color judgments? And what should one
say concerning the meaning that public-language color-words have for them?
Concerning presentational contents: The inverts’ presentational contents
are different from, and are the spectrally inverted counterparts of, those of nor-
mal humans. When an invert sees a green object under suitable illumination, the
object is visually presented as instantiating the property presentational red.9 This
presentational content is nonveridical, because presentational color-properties are
not instantiated in the world. But of course the same is true for the presentational
color-content of the visual experiences of normal humans.
Concerning judgmental contents: For an invert, the phenomenologically consti-
tuted grounding presuppositions governing judgmental color content interact con-
stitutively with externalistic factors in such a way that the invert’s color-judgments
attribute Lockean dispositions to external objects that are inverted relative to the
Lockean dispositions that are attributed by the color-judgments of normal humans.
When an invert sees a green object under suitable illumination, and the invert forms
a judgment that he/she expresses by applying the word ‘green’ to the object, the judg-
ment attributes to the object the property being disposed to reliably cause in oneself, in
suitably favorable viewing circumstances, experiential visual experiences with presen-
tational color-content as-of presentational red. The invert’s judgment is true.
Concerning the meanings of public-language color-words: One can cor-
rectly say either of two things about this, but not in the same breath. On one
hand, there is an important sense in which the words have different meanings
as employed by the inverts from the words’ meanings as employed by normal
humans: viz., the words as used by the inverts express judgments with different
judgmental color-content than judgments that normal humans express with
those words. (The color-ascribing sentences are true given what they mean by the
color words they are employing, but are false given what normal humans mean
by those words.) On the other hand, there is also an important sense in which
the words have the same meanings as employed by the inverts: viz., the publicly
observable linguistic behavior of the inverts fully conforms to that of the wider
human population, insofar as the use of color-words is concerned. (The inverts’
color-ascribing sentences are true, because these sentences conform with compe-
tent practice in deploying color-words; in that respect, the sentences are correctly
9
For expository convenience I am here pretending, as is common in philosophical discussion of
spectral-inversion scenarios, that red and green are spectral inverses of one another.
Phenomenal Intentionality and Secondary Qualities 343
10
More below, in sections 14.8 and 14.9, on truth as correct affirmability under contextually
operative semantic standards.
The Constituents of Perceptual Content and the Role of Perception
344
true, even though neither the computer nor any of its components are red? After
all, doesn’t the computer, when it is in its current monitoring-state, have the very
Lockean-dispositional property that constitutes judgmental redness?
Well, no. Look again at the formulation, in section 14.4, of the proposed dis-
junctive grounding presuppositions that I said govern judgmental color-content.
Clause (ii) says that the pertinent Lockean dispositions are properties possessed by
objects one perceives in one’s ambient environment. The envatted brain doesn’t per-
ceive genuine objects at all in its ambient environment—even though it has system-
atically nonveridical experiences as-of perceiving external objects. In particular, it
does not perceive the computer that is generating its inputs, and it does not perceive
any states of that computer either. Thus, even though the computer, when in certain
states, is indeed disposed to generate in the brain experiences as-of the instantiation
of presentational redness by apparent objects in the envatted brain’s apparent envi-
ronment, this does not suffice to make it the case that the computer, when in that
state, instantiates the property of judgmental redness. In order for the computer to
instantiate that property, it would need to be disposed to reliably cause experiences
as-of presentational redness in an experiencer who visually perceives it (and does so
under suitably favorable viewing circumstances). The envatted brain doesn’t visu-
ally perceive anything—even though it seems to itself to see an external world.
The remarks in the preceding section about the differences between spectral-
inversion scenarios and left-right inversion scenarios suggest the following way of
drawing a general distinction between primary qualities and secondary qualities
within the GTH framework. Primary qualities have these features:
(a) one is directly acquainted with presentational primary qualities in sen-
sory-perceptual experience, via apparent instantiations of them by ap-
parent objects in one’s apparent ambient environment;
(b) this experiential acquaintance constitutes presentational mental refer-
ence to them;
(c) presentational primary qualities are actually instantiated in the world;
and
(d) presentational primary qualities are identical to the corresponding prop-
erties one refers to in judgment (i.e., judgmental primary qualities).
Secondary qualities, on the other hand, have these features:
(a) one is directly acquainted with presentational secondary qualities in
sensory-perceptual experience, via apparent instantiations of them by
apparent objects in one’s apparent ambient environment;
The Constituents of Perceptual Content and the Role of Perception
346
In this section and the next, I will situate the preceding discussion within a wider
philosophical perspective that incorporates certain of my views about matters of
truth and ontology. In a number of writings over the years, sometimes collabora-
tive, I have advocated a general conception of truth that I call contextual seman-
tics (Horgan 2001; Horgan and Timmons 2002; Horgan and Potrč 2006, 2008;
Barnard and Horgan 2006). Some leading ideas are these:
and science. For, such putative objects and properties, if they were included in
the right ontology, would be ontologically vague; and ontological vagueness, we
argue, is impossible.
In sections 14.6 and 14.7, I maintained that although perceptual experience
is nonveridical insofar as its secondary-quality presentational content is con-
cerned, it is typically veridical in its primary-quality presentational content. But
I also would claim that much of the presentational primary-quality content of
perceptual experience is vague. (Within the GTH framework, the rubric of pres-
entational primary qualities, as characterized in section 14.7, includes not just
presentational properties like shape, size, relative position, and the like, but nu-
merous others too, some of which are surely vague—e.g., properties like being
a cup and being a table). Can the veridicality of primary-quality presentational
content be reconciled with the vagueness of presentational primary qualities? My
answer is yes. The key to reconciliation is to embrace the claim that the default
primary-quality presentational content of experience is not only often vague in
various respects, but is also governed by indirect-correspondence semantic stan-
dards. When one’s visual experience presents a cup on a table, the presentational
content of this experience can indeed be veridical—even though putative prop-
erties like cuphood and tablehood are vague and therefore cannot belong to the
right ontology. This vague presentational content bears the contextually pertinent
relation of truth—in this case, a relation of indirect correspondence—to the non-
vague world.11
One final point deserves mention. Thought and discourse often is governed
by indirect-correspondence semantic standards even in metaphysics-oriented
philosophical inquiry. For instance, in urging the difference between
primary-quality presentational content and secondary-quality presentational
content, I myself am happy to say that presentational primary qualities like cup-
hood and cathood are real and also are instantiated in the world, and that sec-
ondary presentational qualities like redness are real too (albeit not instantiated
in the world). In positing such properties, and in calling them real despite their
vagueness, I am deploying property-talk in a manner that itself is governed by
indirect-correspondence standards. By my lights this is a theoretically legitimate
way to talk, in context, even though I also hold that the right ontology cannot
include vague properties. Given austere realism, this kind of double-talk is to
be expected, even in philosophical inquiry about the metaphysics of color. Such
11
One might think that under contextual semantics, the presentational color-content of
visual-perceptual experience turns out to be veridical itself—albeit with its veridicality being a form
of indirect correspondence that does not require presentational color-properties to be instantiated
in the world. But I would maintain that ordinary visual experience does present certain determinate
color-properties as being instantiated in spatiotemporally determinate ways, and thereby has
direct-correspondence truth-conditions insofar as determinate presentational color-properties are
concerned. (Although determinable presentational colors are vague, vagueness does not intrude as
regards the determinate color-content of visual experience.)
Phenomenal Intentionality and Secondary Qualities 349
References
Barnard, R., & Horgan, T. (2006). Truth as mediated correspondence. The Monist, 89, 28–49.
Block, N. (1995). On a confusion about a function of consciousness. Behavioral and Brain
Sciences, 18, 227–247.
Chalmers, D. (2006). Perception and the fall from Eden. In T. Gendler & J. Hawthorne
(Eds.), Perceptual Experience (pp. 49–125). Oxford: Oxford University Press.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.
Farkas, K. (2008). Phenomenal intentionality without compromise. The Monist, 91, 273–293.
Georgalis, N. (2006). The Primacy of the Subjective. Cambridge, MA: MIT Press.
Graham, G., Horgan, T., & Tienson, J. (2007). Consciousness and intentionality. In M.
Velmans & S. Schneider (Eds.), Blackwell Companion to Consciousness (pp. 468–484).
Oxford: Basil Blackwell.
Graham, G., Horgan, T., & Tienson, J. (2009). Phenomenology, intentionality, and the unity
of mind. In B. McLaughlin, A. Beckermann, & S. Walter (Eds.), The Oxford Handbook
of Philosophy of Mind (pp. 512–537). Oxford: Oxford University Press.
Horgan, T. (2001). Contextual semantics and metaphysical realism: Truth as indirect cor-
respondence. In M. Lynch (Ed.), The Nature of Truth: Classic and Contemporary
Perspectives (pp. 67–95). Cambridge, MA: MIT Press.
Horgan, T. (2007). Mental causation and the agent-exclusion problem. Erkenntnis, 67,
183–200.
Horgan, T. (2011). Phenomenal intentionality and the evidential role of perceptual experi-
ence: Comments on Jack Lyons, Perception and Basic Beliefs. Philosophical Studies, 153,
447–455.
Horgan, T., & Graham, G. (2009). Phenomenal intentionality and content determinacy. In
R. Schantz (Ed.), Prospects for Meaning (pp. 321–344). Amsterdam: de Gruyter.
Horgan, T., & Potrč, M. (2006). Abundant truth in an austere world. In M. Lynch & P.
Greenough (Eds.), Truth and Realism: New Essays (pp. 137–167). Oxford: Oxford
University Press.
Horgan, T., & Potrč, M. (2008). Austere Realism: Contextual Semantics Meets Minimal
Ontology. Cambridge, MA: MIT Press.
Horgan, T., & Tienson, J. (2002). The intentionality of phenomenology and the phenom-
enology of intentionality. In D. J. Chalmers (Ed.), Philosophy of Mind: Classical and
Contemporary Readings (pp. 520–533). Oxford and New York: Oxford University Press.
Horgan, T., & Timmons, M. (2002). Conceptual relativity and metaphysical realism.
Philosophical Issues, 12, 74–96.
12
Thanks to Mark Timmons for helpful comments and discussion.
The Constituents of Perceptual Content and the Role of Perception
350
Horgan, T., Tienson, J., & Graham, G. (2004). Phenomenal intentionality and the brain
in a vat. In R. Schantz (ed.), The Externalist Challenge: New Studies on Cognition and
Intentionality (pp. 297–317). Amsterdam: Walter de Gruyter.
Kriegel, U. (2011). The Sources of Intentionality. Oxford: Oxford University Press.
Kriegel, U. (2013). The phenomenal intentionality research program. In U. Kriegel (Ed.),
Phenomenal Intentionality (pp. 1–26). Oxford and New York: Oxford University Press.
Loar, B. (2002). Phenomenal intentionality as the basis of mental content. In D. J. Chalmers
(Ed.), Philosophy of Mind: Classical and Contemporary Readings (pp. 000–000).
Oxford and New York: Oxford University Press.
Lycan, W. (1995). Consciousness and Experience. Cambridge, MA: MIT Press.
Lyons, J. (2009). Perception and Basic Beliefs: Zombies, Modules, and the Problem of the
External World. Oxford and New York: Oxford University Press.
Maund, B. [1997] (2006). Color. Stanford Encyclopedia of Philosophy, http://plato.stanford.
edu/entries/color/.
McGinn, C. (1988). Consciousness and content. Proceedings of the British Academy, 76,
219–239. Reprinted in N. J. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of
Consciousness: Philosophical Debates. Cambridge, MA: MIT Press, 1997.
Pitt, D. (2004). The phenomenology of cognition; or what is it like to think that P? Philosophy
and Phenomenological Research, 69, 1–36.
Siewert, C. P. (1998). The Significance of Consciousness. Princeton: Princeton University
Press.
Strawson, G. (1994). Mental Reality. Cambridge, MA: MIT Press.
Strawson, G. (2008). Real intentionality 3: Why intentionality entails consciousness. In id.,
Real Materialism and Other Essays (pp. 281–305). Oxford: Oxford University Press.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press.
15
Earlier versions of this paper were presented at a workshop at Central European University, and at
talks at the University of Bristol and the University of Cardiff. The authors are very grateful for audiences
for their comments. Research for this paper has received funding from the European Commission’s
Seventh Framework Programme FP7/2007-2013 under grant agreement no. FP7-238128, and from the
project BETEGH09 supported by MAG Zrt. 351
The Constituents of Perceptual Content and the Role of Perception
352
theorizing on the other, is untenable. According to this suggestion, the way we see
the world, the way it appears to us, is already a result of joint work between theory
and observation. This is one of the important questions about the content of expe-
rience: just what is it exactly that perceptual experience presents us?
This question has always interested philosophers: below we shall quote for
example Aristotle’s view on the objects of perceptual experience. However, it is
worth mentioning some factors in and outside philosophy that have shaped the
recent development of the issue. Starting already in the nineteenth century, the
empirical study of perception in cognitive psychology and later, in the twentieth
century, in neuroscience, has become a huge enterprise, and we have now exten-
sive experimental data and a whole array of theories about the psychological and
physiological aspects of perception. We now know that the seemingly simple act of
seeing an object is made possible by a complex operation of the perceptual system,
using both hardwired and learned mechanisms. This gave a new impetus to asking
the question of whether perception can be separated from the rest of our cognitive
operations. This question, in turn, may have far-reaching consequences for our
epistemological theories.
Among the senses, vision has always been in the primary focus of investiga-
tion both in philosophy and psychology. More recently, however, there has been a
growing interest in the other senses, and it seems that focusing on vision may even
have distorted our view of perception. As we mention below, some philosophers
have even considered the view that for example the olfactory sense is not in the
business of presenting the world at all: olfactory experiences have no content, they
are mere sensations, mere modifications of one’s consciousness. We defend a dif-
ferent view in this paper.
1
The boundary between successful misperceptions and hallucinations is probably blurred; it’s not
clear how serious a misperception can be so that we can still say that some object is actually perceived.
But even if the distinction is vague, it is still a distinction, and an important one.
Which Causes of an Experience Are Also Objects of the Experience? 353
on the table, or the surface of one’s retina. Even if the subject of experience knows
that these things participate in the causal chain, and hence she can infer to their
presence or existence from the fact that she has the perceptual experience, it would
still not sound right to say that these are the objects of her experience.
15.3. A Question
Since the possible objects of perceptual verbs show a great variety in a number of
dimensions, philosophers sometimes raise the question of whether some objects
are special in a certain sense. First, the investigation has to be narrowed—if
Which Causes of an Experience Are Also Objects of the Experience? 355
2
Which properties can be represented—or presented, if someone doesn’t want to commit herself
to the idea that perceptual experiences are representational. Hopefully, taking sides on this issue is
orthogonal to the question we are asking. Hence we shall often talk of representation and content, but
what we say should be relevant for those who believe that perceptual experience is not representational
(e.g., Travis 2004).
3
An example of the former, called sometimes ‘sparse’ or ‘conservative’ view, is Tye (1995, pp. 140–
141). Some of the recent defenses of the ‘rich’ or ‘liberal’ view include Siegel (2006) and Bayne (2009).
The Constituents of Perceptual Content and the Role of Perception
356
A further question about the objects of perception concerns the proper objects
of a sense, that is, the objects that can be perceived by that sense only4; for example,
smells can arguably be perceived only by the olfactory sense. Yet another ques-
tion is about the primary objects of perception: these are objects that are always
perceived whenever something is perceived in a sensory modality. For example,
people have argued that whenever we hear something, we always hear a sound—
indeed, if we hear anything, we hear it by or in virtue of hearing a sound. As David
Sanford (1976) makes it clear, though sounds are also proper objects of hearing,
being proper and being primary are different conditions; something could be spe-
cific to a sense yet not involved in every experience through that sense, and some-
thing could be involved in every such experience without being specific to that
sense.
Proper and primary ‘objects’ are sometimes conceived as qualities: for
example, when it is claimed that colors are proper objects of vision. In this case,
‘object’ is not an ontological category, but simply signals that something is per-
ceived in a certain modality. But proper and primary objects need not always be
qualities. A ‘sound’ can be understood as a spatio-temporally located physical
existent.5 Similarly, smells or tastes (or perhaps ‘odors’ and ‘flavors’) can be un-
derstood not as qualities, but as quantities of certain chemicals. If we wanted to
find an equivalent of sounds, odors and flavors in the visual modality, it would be
something like a colored expanse, rather than simply a color.
Developing a theory of the proper and primary objects of perception—if it
can be done—goes some way towards answering our question, but doesn’t go the
whole way. For one thing, saying that a colored expanse is the proper and primary
object of sight doesn’t answer the question of why it is the glass we see, but not the
retina or the glassmaker, given that all three have colored surfaces. Further, we are
interested in all objects of perception, including improper and secondary ones,
and anyone who allows for such objects will still have to single out some elements
in the causal chain. Finally, our question stands even if there are no proper and
primary objects of perception. So the debate about the proper, primary or direct
objects of perception, though relevant to our purposes, is not exactly what we are
interested in.
Our question was: which elements in the causal chain leading to (the final constit-
uent of) perceptual experiences are the objects of experience? That is, where, along
4
See Sanford 1976, p. 190. The question goes back to Aristotle’s theory of ‘proper sensibles’. It
is sometimes thought to have significance for the issue of distinguishing different sensory modalities
from each other; see Nudds (2004).
5
This question is discussed for example in O’Callaghan (2007).
Which Causes of an Experience Are Also Objects of the Experience? 357
6
The phenomenon is general: given that some bodily sensations arguably have an object, the same
observation applies to bodily sensations too. In this paper, we focus mainly on perceptual experiences,
but we discuss for example tactile sensations (i.e., awareness of bodily parts in the tactile mode), since
they have a very intimate connection with external tactile perception.
The Constituents of Perceptual Content and the Role of Perception
358
However, Lycan is reluctant to give up the view that olfactory experiences repre-
sent odors, so he settles for a ‘multi-object’ view: olfactory experiences represent
both roses and odors, indeed they represent the former ‘by’ representing the latter.
The difference between smelling the odor of coffee and smelling a thing like
coffee in the cup illustrates the sense in which olfactory experiences can have dif-
ferent type of objects: different elements are singled out in the causal chain leading
to the experience. This means that there are different conditions for the experience
being successful. If the object of smell is coffee, and there is no coffee in the room,
then any apparent experience of coffee is hallucinatory. But if the object is an odor
which happens to be emitted by coffee, then an experience can be veridical even if
there is no coffee around, but only the smell lingers.7
At first sight, this relation between perceptual error and the content of
experience may suggest a way of finding out what the proper objects of percep-
tion are: those whose presence and properties are responsible specifically for per-
ceptual, rather than other kinds of error. Unfortunately, this observation, while
valid, doesn’t advance the debate at this stage, because we don’t really have direct
intuitions about perceptual error independently of what we take to be the objects
of perception. All parties agree that in the case of mistakenly thinking that one is
smelling a real rose rather than an artificial one, there is an error. But the question
of whether the mistake is perceptual or cognitive, is the same as the question of
what the object of perception is.
7
In fact, the view that smells have no objects at all can be regarded as a ‘limiting case’ of moving
up in the causal chain, closer and closer to the experience (or relevant brain state). Perhaps one
could extend the argument of this paper to show that moving from a pure qualitative condition to an
experience with some content is a process in some ways similar to the process of the object moving
along the causal chain.
Which Causes of an Experience Are Also Objects of the Experience? 359
one of the data we have to account for is that we see the glass in front of us, but we
don’t see the optic nerve, nor the hand that placed the glass here.
There are various notions, like ‘signaling’ or ‘carrying information’ that may
adequately cover the relation between a perceptual experience and any member of
the causal chain that leads to the experience. Perhaps experiences carry informa-
tion about states of the retina, states of the optic nerve; perhaps a visual experi-
ence of the lights being on next door signals that the neighbor is at home. But
the reason why we say that neither the retina, nor the neighbor is the object of
the perception—that we don’t see the retina or the neighbor—is that they are not
phenomenally present in the experience. We don’t seem to see these things when
having the experience—we see (or at least seem to see) simply the lights.
The idea that the objects of perception are phenomenally present in an expe-
rience is meant to be compatible with pretty much any theory of perceptual expe-
rience. Suppose we accept a representationalist theory of perception, which holds
that the nature of a perceptual experience is determined by its representational
properties, which, in turn, determine the accuracy conditions for the experience.
A particular visual experience is of a glass, because the experience is veridical if
the glass—rather than the retina, or the glass-maker—has certain properties. Of
course, we might agree with that. But how did the representationalist know where
to look for the accuracy conditions? Why didn’t she suggest that the experience is
veridical if the retina is in a certain state? The simplest answer, we suggest, is that
the retina doesn’t seem to be the object of the experience—in other words, it’s not
phenomenally present in the experience.8
Appealing to the idea of phenomenal presence is not going to settle all issues
about the objects of perception. First, there may not always be a consensus on what
is phenomenally present in someone’s experience. Each of us has to rely on their
first-person reports, and some people may be better than others in attending to the
phenomenal features of their experiences. In some cases, there may be some un-
certainty at first whether something is phenomenally presented in an experience
or its presence is evoked as a result of some association, and then different kinds
of considerations might be used to decide the question. Besides, as we noted, the
guidance of phenomenal presence is compatible with many, otherwise different
theories, so clearly, not all questions will be solved this way.
But even if there will be unclear cases, it seems obvious that there are clear cases
as well. As we said before, most philosophers will agree that one’s own olfactory bulb
is normally not the object of an olfactory experience. To argue that the olfactory bulb
is the object of an olfactory experience requires significant further work. So it seems
that we do use the condition of phenomenal presence as the default position to
exclude certain objects as candidates for being the object of a perceptual experience.
8
A similar notion of phenomenal presence is used for a similar purpose—i.e., to locate objects
of representation among the many causes of representation—by Galen Strawson in Strawson (2008).
Which Causes of an Experience Are Also Objects of the Experience? 361
perceiving an odor. If that was the case, then we could establish uniformly, for all
cases of the olfactory modality, which elements in the causal chain are objects of
the experience. But our contention is that we cannot establish this.
The most plausible cases for smelling only odors are those of dispersed and
unfamiliar smells. You are walking on a deserted country road and suddenly a
peculiar, faint and slightly stale smell hits your nose. You have no conception of
the origin of the smell; the faintness and staleness is characteristic of a lingering
smell rather than a freshly produced smell. We think that it would be implausible
to say that in this case, any object beyond the odor itself is phenomenally present
in your experience.
A different kind of case is when you feel a feel a familiar smell, clearly asso-
ciated with a source, perhaps even answering to some need or expectation. You
arrive at a friend’s house for brunch after a bad night sleep; you smell freshly
brewed coffee, and you immediately look for the coffee pot. Maybe this is one of
the cases that fits Lycan’s idea of perceiving objects with potential adaptive signif-
icance: what matters for your survival that morning is the coffee, not its odor, and
this may explain its phenomenal salience in your experience. Since the coffee is
presented in your experience, it is an object of the experience.
We have been arguing that on some occasions, the object of an olfactory experi-
ence is an odor, on other occasions, it is (also) the source of the odor. This sup-
ports our claim that within a sensory modality, one cannot fix the object of the
experience for all cases. Our main argument for claiming that the object may vary
from case to case is based on an appeal to the phenomenal presence of the object.
Sometimes it’s the odor, sometimes it’s (also) the source of the odor that is phe-
nomenally present in an experience.
A possible objection is that we don’t in fact smell the coffee in the second case,
but we merely make a quick unconscious inference to the presence of coffee, or that
an immediate association of the smell and the coffee brings the idea of coffee to
mind (in the latter case, we need not say that any inference took place). Coffee is
not the object of the perceptual experience, according to this objection, but of a
separate mental state, probably a judgment. The coffee case then would be similar
to the case of ‘seeing’ that it’s cold outside: we don’t literally see the cold, we only
learn that it is cold on the basis of a visual experience.
According to the view which is behind this objection, the inference or asso-
ciation is unconscious. This is important, because it doesn’t seem that in the
specific cases we have in mind, there is an intermediate step. The idea of coffee
comes to you directly upon entering the house, and immediately prompts a
reaction. (Similarly, to anticipate the discussion of tactile experiences, when your
fingers close around the key in your bag, the process doesn’t seem to consist of two
Which Causes of an Experience Are Also Objects of the Experience? 363
steps: first, the identification of the sensation on your skin, and then matching it
with an awareness of the external object. Instead, you just feel the key.)
We do not want to say, at this point, that when you smell the coffee, you cease
to be aware of the odor, and when you feel the key, you cease to be aware of a con-
dition of your skin. In fact, we’d like to leave this question open. What we want to
say is merely that it doesn’t seem that awareness of the coffee or the key is reached
through an intermediate step. It seems possible to be aware of two different objects
without the phenomenology of indirectness. An analogous case, involving proper-
ties rather than objects, might be the example we mentioned above about seeing
pine trees. Susanna Siegel’s claim that experiences (can come to) represent natural
kind properties does not entail that when they do, they cease to represent simpler
properties like colors or shapes. If Siegel is right, and when developing a recogni-
tional capacity for pine trees, you start to be perceptually aware of something being
a pine tree, this doesn’t mean that you cease to be perceptually aware of the shapes
and colors that determine that something looks like a pine tree. Furthermore, and
that’s the crucial point, the property of being a pine tree that according to Siegel
enters the content of the perceptual experience, does not seem to be represented
in the content of a mental episode as somehow being inferred from, or following
upon, the original perceptual experience of shapes and colors.
Now back to the suggestion that we don’t perceive, but merely unconsciously
infer, or make an association with, the presence of coffee. Our reply is that this
objection is based on a false contrast between perception and the involvement of
unconscious inference or association. If an unconscious inference or association is
involved in a perceptual process, this doesn’t mean that the inference starts from a
perceptual experience and leads to a separate mental state. It’s very much possible
that the unconscious process precedes the formation of the perceptual experience
itself.
There is overwhelming evidence that our perceptual experiences are results of a
substantial amount of unconscious work performed by our perceptual system. This
is especially well-documented in the case of visual perception.9 Take for example the
hollow-face illusion (Gregory 1973, pp. 49–96). The stimuli that reach the sensory
organ are compatible with both a concave and a convex surface under different illu-
mination conditions, and yet the resulting experience is always of a convex surface.
It seems that the best way to explain this fact is to assume that our perceptual system
plays an active part in producing a perceptual experience. One idea is that because
of previous experiences of faces which were all convex, the visual system ‘judges’ it
to be much more likely that the present experience is of a convex surface, and hence
9
There is a debate about how this ‘work’ is best characterized: for example, as an unconscious
inference following Bayesian principles, or merely as a process conforming to various laws. This debate
is not our concern here: we simply want to rely on the fact that the content of perceptual experiences
is not entirely determined by the stimuli on our perceptual organs, but requires the active contribution
of the perceptual system as well.
The Constituents of Perceptual Content and the Role of Perception
364
infers from the ambiguous stimuli to the presence of a convex surface. Of course,
if there is inference here, it must be unconscious, and some people even object to
saying that the visual system can do anything like judging or inferring, and some
other expression should be used. Be that as it may, the unconscious inference cannot
be understood as starting from a perceptual experience and resulting in a separate
mental state. Instead, the inference precedes the experience itself: there is no doubt
that we have a genuine visual perception of the face as convex.
The relevant examples vary a great deal, but the upshot is the following. Just
because unconscious association or interpretation, based on previous experiences,
is involved in a perceptual process, this doesn’t mean that there is first a percep-
tual experience, and then the interpretation produces a separate judgment. The
unconscious interpretation may well be part of the formation of the perceptual
experience itself. We need an explanation of why unconscious interpretation in
the coffee case is different from unconscious interpretation in the case of the hol-
low face illusion; until the difference is demonstrated and shown to have explan-
atory advantages over our proposal, phenomenal presence remains the best guide
to what the object of perception is.
We did say above that phenomenal presence may not settle all issues about
the objects of perception. We do insist, however, that it is our primary guide, and
if phenomenal presence singles out something as an object of a perceptual experi-
ence, we need further, and decisive reasons to dismiss the testimony of phenom-
enal presence. On this basis, coffee qualifies as an object, and the fact that this is
the result of an unconscious interpretation does not disqualify its status as object
of perceptual experience.
One question may remain, however: what is then the difference between the
coffee case and the case of ‘seeing’ that it’s cold? Why doesn’t the second qualify
as a genuine case of perceiving temperature? We shall answer this question in the
next section.
In this section and the next we shall illustrate how the object of perception can
move along the causal chain in the case of experiences in the tactile modality.
Philosophers have expressed different views about what should be regarded as the
object of touch. On the most ‘minimal’ version, the object is some part of the
skin: in touching something, we feel a pressure on our skin. A possible variation
on this view would also include the muscles communicating the magnitude of
pressure.
Perhaps some people don’t regard the skin or the muscles as proper ‘objects’
of experience; they may argue that these cases are better described as mere sensa-
tions, that is, sensory experiences without a perceptual object. We do not need
to take issue with this, because this way of looking at things still fits our general
Which Causes of an Experience Are Also Objects of the Experience? 365
thesis: the object of experience is not fixed for a certain sensory modality. However,
what matters here is that tactile objects and bodily parts are different links identi-
fied along the same causal chains that result in perceptual experiences. For the sake
of simplicity, we will continue talking of ‘tactile experiences’ as covering both cases
of bodily and tactile perception.
If the skin and what’s within the skin are the only objects of a tactile experi-
ence, anything that extends further than the skin must be the object of a separate
mental state. This is the view held by Fred Dretske, for example:
. . . even if we can tell the shape of things by touch (i.e., feel that it is square),
I do not think this shows we feel the object’s shape. What we feel when we
tell the shape of objects in this way is pressure and (if this is really different)
texture. It is differences in pressure we feel as we move our hand around the
object that (together with what we know about how we are moving our hand)
tell us what shape an object is. [ . . . ] If this is to count as feeling shape, then
wine connoisseurs must be tasting colors. (Dretske 2000, pp. 458–459)10
To be precise, Dretske is talking here about the question of which properties, rather
than which things are perceived. But there is an obvious connection between these
questions: perceiving a thing is perceiving it as being in some way, that is, as exem-
plifying some properties. Presumably, the properties that enter the content of an
experience are the ones exemplified by the things we perceive. Hence if shape and
texture are not something we literally perceive when having a tactile experience,
then the object that has the shape and texture is not the object of the experience.
Dretske’s claim about tactile experiences falls within our discussion from the
previous section and so all the considerations and conclusions there apply to the
present case. That is, since objects outside the skin often have an immediate phe-
nomenal presence in tactile experiences, on our account, they qualify as objects
of perception.
We agree that it would be odd to grant that wine connoisseurs taste colors,
but we deny that tactile experience of extrabodily objects is analogous to ‘tasting
colors’. First, even for expert wine tasters, it is questionable how much the color of
the wine is phenomenally present in a gustatory experience. Take one of the more
easily identifiable grape varieties, for example pinot noir. It seems conceivable that
similarly to the example of the pine tree, someone can develop a reliable recog-
nitional capacity for pinot noir through tasting, and being pinot noir becomes
part of the content of their gustatory experience. Pinot noir is red, so one can
readily identify the color of the wine on the basis of the flavor. But this isn’t the
same as having redness phenomenally present in one’s gustatory experience: the
10
That is, the claim that extrabodily objects are not the objects of tactile experiences is Dretske’s
view only in case he regards feeling texture as “really different” from feeling pressure. We’ll assume here
that this is his view while the reader should keep in mind the qualification ‘unless texture is different
from pressure’.
The Constituents of Perceptual Content and the Role of Perception
366
identification of the pinot noir flavor does seem to be an intermediary step on the
way to the identification of color.
Why is there an intermediary step? Because it seems that there is no phenom-
enally unified category of the flavor of red (or the flavor of white, for that matter).
Red wines of different grape varieties or by different makers taste very different,
and then we haven’t even considered other red beverages, like raspberry juice or
rooibos tea. This is why the identification of color on the basis of flavor seems to us
indirect: the flavor that indicates red is too diverse to make it possible for red to be
presented phenomenally in a gustatory experience.
To come back to the question we left unanswered at the end of the previous
section: the same holds for our earlier example of ‘seeing’ that it’s cold. The visual
appearance of it’s being cold is just too diverse to form a category that could be
directly phenomenally present in experience. We learn that it’s cold when we see
snow, ice, people wearing thick coats, their having red noses and blue lips, their
breath steaming in the cold air, and so on. There is no such thing as phenomenal
category of ‘the look of cold’.
But the case of felt shapes, textures and temperatures is different. In these
cases, there is a phenomenally robust quality that is present in tactile experi-
ences. That is why we find it overwhelmingly plausible that in these cases, the
thing which has the shape or texture can be a genuine object of the perceptual
experience.
As before, to show that the object of experience moves along the causal chain,
we need to present instances both of feeling only the condition of the skin or bodily
part, and other instances when we feel (also) the object that touches the skin. We
have already mentioned a couple of relevant examples. The most plausible cases of
feeling only a bodily part are cases of unexpected, fleeting and unfamiliar tactile
sensations. If the sensation is unpleasant—for example a sort of sting verging on
the slightly painful—then it’s all the more plausible that the center stage of one’s
phenomenal awareness in the experience is taken up by the skin, with no room left
for the offending object that caused the sensation.
How different this is from a purposeful search among familiar objects. The
earlier example was searching for a key; another one may be finding the utterly
familiar location of the switch of your bedside lamp in the dark. It seems to us that
as soon as one’s finger hits on the switch, we do genuinely feel its presence.
11
Tactile visual substitution system (TVSS) is but one of the many sensory substitution systems
that have been developed and tested, and other cases seem to confirm that the subjects using the
systems come to perceive new objects previously not experienced in the modality (see Bach-y-Rita
1996).
The Constituents of Perceptual Content and the Role of Perception
368
Both Merleau-Ponty’s description and the reports about using sensory substi-
tution systems raise a question that we already touched upon in passing: when a
more remote link in the causal chain is added as an object, do we cease to perceive
the more proximate links? Going back to the case of olfaction, the question would
be whether we cease to perceive an odor, once its source becomes the object of the
olfactory experience. Or similarly for tactile modality—do we cease to perceive
what happens to the palms of our hands once we engage in an active explora-
tory touch of an extrabodily object? We cannot settle the question here, but some
possible answers were already mentioned. Lycan’s layered-representation view of
olfaction holds that we always perceive the odor as well as the source of the odor.
Merleau-Ponty has a different view in the case of touch: once the stick becomes a
familiar tool, it ceases to be an object of perception, and the only things felt are the
explored surfaces.
15.11. Conclusion
References
Aristotle (1995). On the soul. Trans. by J. A. Smith. In Jonathan Barnes (Ed.), The Complete
Works of Aristotle (pp. 641–692). Princeton: Princeton University Press.
Bach-Y-Rita, Paul (1996). Sensory substitution and qualia. In Alva Noë & Evan Thompson
(Eds.), Vision and Mind (pp. 497–514). Cambridge, MA: MIT Press (reprinted in 2002).
Bayne, Tim (2009). Perceptual experience and the reach of phenomenal content. The
Philosophical Quarterly, 59, 385–404.
Dretske, Fred (2000). Reply to Lopes. Philosophy and Phenomenological Research, 60,
455–459.
Fodor, Jerry A. (1987). Psychosemantics. Cambridge, MA: MIT Press.
Gregory, Richard (1973). Illusion in Nature and Art. London: Duckworth.
Lycan, William G. (1996). Layered perceptual representation. Philosophical Issues, 7, 81–100.
Merleau-Ponty, Maurice (1945). Phenomenology of Perception. Trans. by Colin Smith.
London: Routledge & Kegan Paul, 1962; translation revised by Forrest Williams, 1981;
reprinted, 2002.
Which Causes of an Experience Are Also Objects of the Experience? 369
Nudds, Matthew (2004). The significance of the senses. Proceedings of Aristotelian Society,
New Series, 104, 31–51.
O’Callaghan, Casey (2007). Sounds: A Philosophical Theory. Oxford: Oxford University
Press.
Peacocke, Christopher (1983). Sense and Content: Experience, Thought and their Relations.
Oxford: Oxford University Press.
Sanford, David H. (1976). The primary objects of perception. Mind, 85, 189–208.
Siegel, Susanna (2006). Which properties are represented in perception? In Tamar Gendler
Szabo & John Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford
University Press.
Siegel, Susanna (2013). The contents of perception. In Edward N. Zalta (Ed.), The Stanford
Encyclopedia of Philosophy. \ http://plato.stanford.edu/archives/spr2013/entries/
perception-contents/.
Strawson, Galen (2008). Real intentionality 3. In id., Real Materialism and Other Essays (pp.
281–305). Oxdford: Oxford University Press.
Travis, Charles (2004). The silence of the senses. Mind, 113(449), 57–94.
Tye, Michael (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press.
INDEX
371
Index
372
Experiences: auditory, 274, 278, 356; relationalism, 211; and causal relations, 352;
belief-independence of, 82n14, 85; bipolarity and content view, 111–113; and disjunctivism,
of, 112–113, 126; and justification, 85–91; 179; gappy content view, 24, 294–299;
modularity of perceptual experience, 28–30, indexicals view, 302–303; misperception vs.,
80n7, 82n14, 315–316; objects of, 352–354; 352n1; and naïve realism, 226–227; negative
olfactory, 31–32, 275, 275n19, 278–279, 356–358; account of, 226–227; phenomenally different,
presentational sensory experience, 107–108; 305, 305n24; positive account of, 226, 227n10;
and reasons, 81–85; soliciting, 55; tactile, possible worlds conception, 299–302; and
274–275, 278, 364–366; token, 190–191, 303 presentation, 253; and representationalism,
Experiential representation, 81n10 304–306; and strong content view, 3–4
Externalistic intentionality, 332–334, 340–341 Haptic experiences, 275
Harman, Gilbert, 111
Facts, perception of, 128–129 HD (habit-discontinuity) thesis, 63
Farkas, Katalin, 31–32, 351 Heck, Richard, 78–79, 80
Farrenikova, A., 312n3 Heim, Irene, 25
Feature-placing structures, 271–273, 272n13, Hellie, Benj, 22, 242
277–279 Hering Illusion, 115, 117
Felt solicitation, 55 High-level properties, 18, 21, 28–30, 311–313
Ferrante, D., 324 Hollow face illusion, 43, 363
Fish, Bill, 229 Horgan, Terry, 30–31, 329, 331, 331n4, 335n5.
Flavor, 275n17, 279 See also Graham-Horgan-Tienson (GTH)
Fodor, Jerry, 312, 315, 359 account
Folk-psychology, 82 Hume, David, 267n3, 269, 281–282
Force-selection, 160, 170 Hyperintensionality, 10–11, 92–93, 99n38
Frege, Gottlob, 145, 146, 154, 172–173, 175, 203 Hyper-intrinsic approach, 341
Fundamentality, 201
Further explored objects, 366–368 Iconic representation, 81n10
Illusions, 11–12, 112–136; argument from,
Gappy content, 24, 293, 294–299, 303 313n5; and austere relationalism, 216–217;
General compositionality, 100n38 and beliefs, 80n7; and causal relations,
Generalizations, 144–147, 161 352; and disjunctivism, 179; and dorsal
Gerbino, I., 324 perception, 43; and naïve realism, 225–226;
Gestalt psychology, 271 and presentation, 253; and redundancy of
Gibson, J. J., 51, 275 EX-ings, 134–135; and shifting aspects, 153
Glüer, Kathrin, 9–11, 76 Image content, 24–26, 265–290; and belief, 268–
Gombrich, Ernst, 315 271; deflationary views, 284–287; elements
Graff, D., 184, 189 of, 271–277; limitations, 279–280, 281–282;
Graham, George, 331. See also and naïve realism, 287–288; part-whole
Graham-Horgan-Tienson (GTH) account composition, 277; and perception, 266–268;
Graham-Horgan-Tienson (GTH) account, predicative feature-placing, 271–273, 277–
329–350; and austere realism, 347–349; and 279; property coding, 276; and propositional
colors, 334–340; and contextual semantics, content, 267–268; shared, 280–281, 283–284;
346–347; and envatment, 341–345; and spatiotemporal connectedness, 273–275
left-right inversion, 341–345; phenomenal Immediacy, 82n14
intentionality vs. externalistic intentionality, Imperatival contents, 71n26
332–334; primary/secondary distinction, Inattentional blindness, 63
345–346; role of, 340–341; and spectrum Indeterminacy, 204
inversion, 341–345 Indexicals, 25, 302–303
Grice, P., 314n8 Indiscriminability, 15, 180, 181–184, 191–194
Grounding presuppositions, 30, 205, 334 Information-carrying, 44–45
Gupta, Anil, 285, 286, 287n26 Instancing, 145, 146
Instantiation, 183
Habit-discontinuity (HD) thesis, 63 Intelligibility, 250
Habit formation, 62 Intentionality, 59, 73, 200n2, 224, 304. See also
Hallucinations, 24–25, 291–308; and austere Representation
Index
374
Visual perception, 116. See also Visual Wayward causal chains, 112
representation Weak content views, 1–5
Visual phenomenology, 279 Weak modularity thesis, 315
Visual representation, 311–328; and Westerståhl, Dag, 100n38
aspect-perception, 321–322, 323–326; Williamson, T., 179, 184, 189
conservative views on, 311–312; disputes over, Wittgenstein, Ludwig, 138, 172, 311, 321, 322
313–315; layering views, 319–321; modularity Wollheim, Richard, 322
thesis, 315–316; sensory qualities, 322–323;
Siegel’s view on, 316–319 Xu, Fei, 271n11