Cognitive Penetration of Colour Experience: Rethinking the Issue in Light of an Indirect Mechanism

Abstract

Can the phenomenal character of perceptual experience be altered by the states of one's cognitive system, for example, one's thoughts or beliefs? If one thinks that this can happen (at least in certain ways that are identified in the paper) then one thinks that there can be cognitive penetration of perceptual experience; otherwise, one thinks that perceptual experience is cognitively impenetrable. I claim that there is one alleged case of cognitive penetration that cannot be explained away by the standard strategies one can typically use to explain away alleged cases. The case is one in which it seems subjects' beliefs about the typical colour of objects affects their colour experience. I propose a two-step mechanism of indirect cognitive penetration that explains how cognitive penetration may occur. I show that there is independent evidence that each step in this process can occur. I suspect that people who are opposed to the idea that perceptual experience is cognitively penetrable will be less opposed to the idea when they come to consider this indirect mechanism and that those who are generally sympathetic to the idea of cognitive penetrability will welcome the elucidation of this plausible mechanism.

Full text

Cognitive Penetration of Colour
Experience: Rethinking the Issue in
Light of an Indirect Mechanism
fiona macpherson
University of Glasgow
Can the phenomenal character of perceptual experience be altered by the states of
one’s cognitive system, for example, one’s thoughts or beliefs? If one thinks that
this can happen (at least in certain ways that are identified in the paper) then one
thinks that there can be cognitive penetration of perceptual experience; otherwise,
one thinks that perceptual experience is cognitively impenetrable. I claim that there
is one alleged case of cognitive penetration that cannot be explained away by the
standard strategies one can typically use to explain away alleged cases. The case is
one in which it seems subjects’ beliefs about the typical colour of objects affects
their colour experience. I propose a two-step mechanism of indirect cognitive
penetration that explains how cognitive penetration may occur. I show that there
is independent evidence that each step in this process can occur. I suspect that
people who are opposed to the idea that perceptual experience is cognitively pene-
trable will be less opposed to the idea when they come to consider this indirect
mechanism and that those who are generally sympathetic to the idea of cognitive
penetrability will welcome the elucidation of this plausible mechanism.
1. Introduction
Can the phenomenal character of perceptual experience be altered by
the states of one’s cognitive system, for example, one’s thoughts or
beliefs? If one thinks that this can happen (at least in certain ways that
will be identified below) then one thinks that there can be cognitive
penetration of perceptual experience; otherwise, one thinks that percep-
tual experience is cognitively impenetrable. Prior to encountering one
particular counterexample, I found the cognitive impenetrability thesis
rather plausible. Although there are myriad alleged counterexamples to
the cognitive impenetrability thesis, there are a number of good strate-
gies to deal with them. Most of the alleged counterexamples involve
cases in which it is claimed there are two experiences with different
phenomenal characters and the reason for this difference is best
24 FIONA MACPHERSON
Philosophy and Phenomenological Research
Vol. LXXXIV No. 1, January 2012
!2011 Philosophy and Phenomenological Research, LLC
Philosophy and
Phenomenological Research

explained by a difference in the propositional attitudes of two subjects
who each have one of the experiences (or by a difference in the propo-
sitional attitudes of one subject who has one experience and one set of
propositional attitudes at one time and the other experience and
another set of propositional attitudes at a later time). One of the strate-
gies to explain away these cases is to claim that there is no difference
in the phenomenal character of the experiences in a given case – there
is merely a difference (or change) in the judgments made on the basis
of experiences with the same phenomenal character. Another strategy is
to claim that, while the experiences differ in phenomenal character, this
change hasn’t come about because a state of the cognitive system has
penetrated experience, but on account of a change in perceptual pro-
cessing – such as a change in perceptual attention or in eye movement.
However, there is one potential counterexample to the cognitive
impenetrability thesis that is not persuasively explained away by the
usual strategies. The case is one in which it seems subjects’ beliefs
about the typical colour of objects affects how they experience the
colour of objects (Delk and Fillenbaum (1965)). After explaining why
I think this case is hard to explain away, I propose a mechanism of
indirect cognitive penetration that explains how cognitive penetration
may occur. The mechanism is indirect because the penetrating belief
causes some non-perceptual state with phenomenal character to come
into existence or causes some process to occur that typically would
yield a non-perceptual state with phenomenal character, and then that
non-perceptual state or process affects, interacts with, or modifies, the
phenomenal character of the perceptual state. I show that there is
independent evidence that each step in this process can occur. I suspect
that people who are opposed to the idea that perceptual experience
is cognitively penetrable will be less opposed to the idea when they
come to consider this indirect mechanism. And I hope that those
who are generally sympathetic to the idea of cognitive penetrability will
welcome the elucidation of this plausible mechanism.
The structure of this paper is as follows. In section two, I will spell
out in detail the thesis that perceptual experience is cognitively impene-
trable and give reasons why people think that it is true. In section three,
I will provide examples of challenges to the thesis and elucidate the
ways in which one can try to reject them. In section four, I will outline
the case concerning colour that I believe is hardest for the supporter
of the cognitive impenetrability thesis to deal with. In section five, I out-
line the indirect penetrative mechanism that I believe may explain the
cognitive penetration that seems to take place in the problematic case.
I also give reasons to think that this mechanism could be realised in
humans. The first stage of the mechanism happens frequently when
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 25

imagining, dreaming and hallucinating. We have evidence that second
stage of the mechanism occurs in humans because we have evidence that
the Perky effect occurs and evidence from the nature of some dreams
and hallucinations. Putting together these two stages yields my proposal
for the indirect mechanism. Finally, I speculate on other perceptual phe-
nomena that may be explained by the same mechanism.
2. The Nature of the Cognitive Impenetrability Claim
The claim that perception is not cognitively penetrable is a thesis that
has received much attention in philosophy and psychology. In fact, two
different forms of the claim have been made. One form of the claim
has been recently advocated at length by Pylyshyn (1999). He argues
that the content of the early visual system cannot be altered by the
higher-level, cognitive system of the brain. He defines the early visual
system functionally, as a system that takes attentionally modulated
signals from the eyes (and perhaps some information from other
sensory modalities) as inputs, and produces shape, size and colour
representations – representations of visual properties – as output. These
representations are then categorised and identified by the cognitive
system making use of memory, knowledge and judgment. Pylyshyn
holds that the content of early vision is cognitively impenetrable, that
is, cannot be altered in virtue of the content of the cognitive system (as
opposed to other features of the cognitive system):
if a system is cognitively penetrable then the function it computes is
sensitive, in a semantically coherent way, to the organism’s goals and
beliefs, that is, it can be altered in a way that bears some logical
relation to what the person knows (1999: 343)
This condition is required to ensure that cases of the following kind do
not count as cases of cognitive penetration: I believe that today is the
day of an important exam. This belief causes stress and brings on a
migraine. The migraine causes disturbances to my vision and I now
experience flashing lights at the periphery of my visual field in addition
to experiencing the scene in front of me. In this case, my visual experi-
ence has been altered on account of my belief but there is no intelligi-
ble connection between the content of the belief – my exam is today –
and the content of my visual experience – the apparent flashing lights –
even though we understand why the belief caused this experience
(namely, stress can cause migraines and these can cause visual distur-
bances).
Pylyshyn also explicitly denies that the output of the early visual
system should be thought of as being or determining one’s visual
26 FIONA MACPHERSON

experience or, put differently, that the content of the output of the
early visual system should be thought of as the content of visual experi-
ence. (The content of a system or of an experience is specified by saying
what is represented by that system or experience.) Pylyshyn’s thesis is,
therefore, not about perceptual experience but about the subpersonal
representational outputs of brain processing mechanisms.
1
However, in philosophy and in psychology, the claim that perception
is cognitively impenetrable is usually taken to be one about perceptual
experience. The claim is that the phenomenal character of perceptual
experience, that is, what it is like to undergo a perceptual experience,
cannot be altered in certain ways in virtue of the content of states of
the cognitive system – states such as thoughts, beliefs, desires, and
other propositional attitudes – that one may possess.
2,3
Moreover, on
the innocuous assumption that I will make for the purposes of this
1
Nonetheless, paradoxically given his insistence that his claim is not about perceptual
experience, in that article Pylyshyn offers evidence for his claim that relies on facts
concerning the nature of the phenomenal character perceptual experience. For
example, when he cites the fact that visual illusions persist, despite one’s knowledge
of the illusion, he is referring to the constant, unchanging nature of perceptual expe-
rience in the face of gaining relevant new beliefs at odds with the experience (1999:
344).
2
The phrase ‘‘what it is like’’ was introduced into philosophy as a way to capture
phenomenal character by Nagel (1974).
3
When thinking about what states of one’s cognitive system are relevant to cognitive
penetration one is naturally drawn towards thinking about the different beliefs and
desires that one has. However, there might be other states of one’s cognitive system
that cognitively penetrate. One example comes from noting that the state of one’s
cognitive system is determined, in part, by which concepts one possesses. If the pos-
session of a concept affected one’s perceptual experience then that state of concept
possession would be one that cognitively penetrated one’s experience. (Of course
some people think that which concepts one has is a matter closely tied to the issue
of which beliefs and desires one has – indeed some even go so far as to claim that
one determines the other.) This type of cognitive penetration would be a form of
‘‘category perception’’. We know that people with different concepts judge objects
in the world differently. One person may judge that an object falls under a category
that another person does not on account of their possession of different concepts.
For example, it has been claimed that which colour concepts one possesses can
affect one’s judgements (and the speed of one’s judgements) about colour (Winawer
et al. 2007) and possession of concepts pertaining to types of animal based on the
patterns of marking on their hide affects one’s judgements (and the speed of one’s
judgments) about types of animal (Goldstein and Davidoff2008). If the differing
judgments are reflective of differing perceptual experiences then these would be
examples of cognitive penetration. Another example of a state of the cognitive sys-
tem that may affect one’s experience is the cognitive system being primed so that
certain concepts are likely to be triggered or activated. (In such a case one is in a
cognitive state such that one is likely to form beliefs or desires involving those con-
cepts, perhaps as opposed to other beliefs or desires.) If being in such a state
affected one’s perceptual experience then that state would be one that cognitively
penetrated perceptual experience.
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 27

paper, namely, that there is no change in the content of a perceptual
experience without a change in the phenomenal character of that expe-
rience and vice versa, then the claim will be that the phenomenal
character of perceptual experience and the content of perceptual experi-
ence cannot be altered in certain ways in virtue of the content of states
of the cognitive system.
This rough statement of the cognitive impenetrability claim needs
to be made more precise. We all know that which beliefs or desires
one has can alter what one chooses to perceive because they can moti-
vate the turning of one’s head, eyes and body – and the position of
these will, of course, affect what perceptual experience one has. Simi-
larly, if one is perceiving a stationary unchanging scene and the condi-
tions for perceiving remain unchanged and one doesn’t move one’s
body or eyes at all, then perhaps one can have different experiences
just in virtue of attending to different parts of the scene. And, of
course, the location one attends to in a scene can be influenced by
one’s beliefs and desires. This type of spatial attention is often
thought of as a subtle form choosing what to perceive that, instead of
involving any movements of the body, just involves a shift in spatial
attention.
4
Although both of these cases involve states of the cognitive
system affecting which perceptual experiences one has, they are not
typically taken to be cases of cognitive penetration. A case of cogni-
tive penetration is a case in which that which is perceived, the viewing
conditions, and the state of the sensory organ are held fixed. One way
to capture this is to imagine the proximal stimulus that affects the
sensory organ (light in the case of the eye, sound waves in the case of
the ear, etc.) and the condition of the sensory organ remaining exactly
fixed. In addition, the location of one’s attentional focus is held fixed.
If it is possible for two subjects in these conditions to have different
perceptual experiences (different in respect of phenomenal character
and content) on account of the differing states of their cognitive
systems, or if it is possible for one subject at different times in these
conditions to have different experiences on account of the difference
between the states of their cognitive systems at those times, then
cognitive penetration is possible. In addition, as we have already seen
by looking at the migraine case, it needs to be added that there have
to be some links between the content of the cognitive state and the
content of the perceptual state that is affected of a nature such that
the effect on the content of the perceptual experience is made intelligi-
ble, or in some very minimal sense rational, in light of the content of
the cognitive state.
4
See Pylyshyn (1999) and Raftopoulos (2001).
28 FIONA MACPHERSON

Thus, perceptual experience is cognitively impenetrable if it is not
possible for two subjects (or one subject at different times) to have two
different experiences on account of a difference in their cognitive sys-
tems which makes this difference intelligible when certain facts about
the case are held fixed, namely, the nature of the proximal stimulus on
the sensory organ, the state of the sensory organ, and the location of
attentional focus of the subject. It is this version of the cognitive thesis
that is under investigation in this paper.
The cognitive impenetrability thesis about experience is often held in
conjunction with a thesis like Pylyshyn’s about early visual processing.
The thought is that the phenomenology of perceptual experience is
determined by the output of some specifiable early visual system. Thus,
frequently, claims about early vision and claims about perceptual expe-
rience go hand-in hand.
For example, one way the cognitive impenetrability thesis is stated is
by claiming that the visual modalities – vision, audition, touch, and the
like – are modular. The idea is that the processing that goes on in the
brain in, say, vision is specialised and isolated – that is, is information-
ally encapsulated – from the other perceptual modalities and from the
more general cognitive system. Information from a sensory organ
enters the brain and is processed discretely in the perceptual mechanism
associated with that organ and the output of this mechanism serves as
input to the cognitive system. There is no top-down processing, that is
no feedback, from the cognitive system into the perceptual modules
(although there may be top-down processing within the perceptual sys-
tem).
5
(It is further claimed that the perceptual modules process sensory
input on the basis of certain assumptions or information about the
world built into the modules. These assumptions are fixed and them-
selves cognitively impenetrable.) Unlike Pylyshyn’s view, the thesis is
not taken to be one merely about subpersonal brain processing and
architecture. Rather, it is assumed that the output of the sensory mod-
ules determines, or is identical with, the content and phenomenal char-
acter of perceptual experience. This view is promoted at length by
Fodor (1983). However, it is only the claim that perceptual experience
is cognitively penetrated that is the one at issue in this paper.
One major motivation for the view that perception is cognitively
impenetrable is the persistence of visual illusions.
6
In such illusions,
5
This is a brief sketch of what is meant when it is said that a system is modular. See
Fodor (1983) for much greater detail. Much debate has taken place in the literature
about how one should spell out modularity. This debate is beyond the scope of this
paper – but harmlessly so. The details of the nature of modularity are tangential to
the arguments here.
6
See Fodor (1983).
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 29

one’s visual experience presents the world as being one way, and con-
tinues to do so, even when one knows that the world is not that way.
For example, in the Mu
¨ller-Lyer illusion one’s visual experience is such
that the horizontal lines seem to be of different lengths, and this experi-
ence persists even when one knows that the lines are the same length.
Of course the existence of one example where some specific belief
about the world does not impact upon perceptual experience does not
prove the thesis that perceptual experiences are never cognitively pene-
trated. At best, it establishes that the content of perceptual experience is
not responsive to the content of belief in the same way as, or to the
same extent as, our other beliefs are. For example, at least on pain of
irrationality but perhaps tout court, our beliefs must have consistent
content – certainly this seems to be true in the case of conscious beliefs
with explicit contradictory content. But in the case of persisting
illusions, in which there is no plausible reason to believe that one is
irrational, the content of one’s experience and the content of one’s con-
scious belief is not consistent. So the dispute about whether perceptual
experience is cognitively impenetrable or not tends to proceed on the
basis of cases. Those who are against cognitive impenetrability present
examples in which it looks as if subjects’ propositional attitudes affect
the nature of the phenomenal character and content of their experience.
Those who defend the thesis try to explain away these examples.
There are several disputes in current philosophy that are closely
related to the question of whether perceptual experience is cognitively
penetrable. One such dispute is whether observation is ‘‘theory-neutral’’.
Let me explain.
The claim that observation is theory-neutral, as advocated by Fodor
(1984), is the claim that there are a core group of beliefs – those formed
directly on the basis of perceptual experience – and those beliefs are not
Müller-Lyer Figure
30 FIONA MACPHERSON

affected by one’s other beliefs. Arguing for the opposing side, Churchland
(1979 and 1988) claims that there are no perceptual beliefs unaffected by
one’s other, theoretical, beliefs. All beliefs are theory-laden.
A natural position for one such as Churchland to adopt would be
that experience is thoroughly cognitively penetrated, for this would
explain why perceptual beliefs are also affected by other beliefs. And
indeed Churchland does adopt this position. But it would be open to
Churchland not to do so. He could hold that perceptual experiences
are cognitively impenetrable but all beliefs that we form on the basis of
such experiences are dependent on our other beliefs and theoretical
commitments. Indeed, Churchland seems amenable to thinking that
there is a core element of perceptual experience, which he calls percep-
tual sensation, that is cognitively impenetrable, however, he thinks
perceptual experience consists in more than this sensational nub. In
short, if one thinks that there are no theory-neutral beliefs it is open to
one to hold either that perceptual experience is cognitively penetrable
or cognitively impenetrable.
However, a proponent, such as Fodor, of the theory-neutrality of
observation beliefs must be committed to the cognitive impenetrability
thesis. For if one’s existing beliefs can affect which perceptual experi-
ences one has then they will indirectly affect which perceptual beliefs
one has, because different experiences will cause different perceptual
beliefs. Thus, the cognitive impenetrability of perceptual experience is
a necessary but not sufficient condition for the truth of the theory-
neutrality of observation but, as we saw in the previous paragraph, it is
neither a necessary nor a sufficient condition for the truth of the
theory-ladenness of observation.
Another dispute in current philosophy that is closely related to the
question of whether perceptual experience is cognitively penetrable is
the debate concerning the admissible contents of experience. This
dispute concerns which properties are represented in perceptual experi-
ence. At present, the debate in the literature focuses on the content of
human visual experience. One view is that our visual experiences have
only ‘‘low-level’’ content, such as content about shape, colour, position
in space, and perhaps object-hood. On this view, your experience might
represent an object of a certain size, shape and colour in front of you.
Of course, if the occasion arose that the object in front of you was
your brother, and you recognised that, it would be natural to think
that that object also looked to you like a human being and your
brother. But on this view, that thought is to be cashed out by saying
that on the basis of your experience that only represented low-level
properties, such as size, shape and colour, you judged that it was a
human being and your brother. Forming this judgment might be
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 31

unconscious, spontaneous, unbidden, and take place incredibly quickly,
but it is a judgment nonetheless. The representation of a human being
and your brother is not part of the content of the visual experience;
rather, it is the content of some cognitive state formed on the basis of
the experience together with one’s memory and background knowledge.
Thus, if you had an experience that represented an object of a certain
size shape and colour, and at first you didn’t realise that it was your
brother but then you came to realise that it was your brother, this view
would say that your visual experience did not change in content or in
phenomenal character. The change in your mental life is a change in
non-perceptual cognitive states, such as beliefs or judgments.
7
The opposing view is that visual experiences have ‘‘high-level’’
content, such as content about natural kinds (for example about
human beings or tigers or pine trees), content about causation, about
particular individuals (such as your brother, as opposed to something
with the look of your brother) and other content that outstrips,
roughly, properties pertaining to shape, size, position, colour and
object-hood. On this view, if you had an experience that represented an
object with certain low-level properties, and at first you didn’t realise
that what you was seeing was your brother but then you came to
realise that it was your brother, then an acceptable explanation might
be that your visual experience did change in content and in phenome-
nal character. Your experience came to represent that a human being
that was your brother was present. Thus, on this view, the change in
your mental life can be a change in the character or content of your
perceptual experience.
People who subscribe to the existence of high-level content in visual
experience are likely to reject cognitive impenetrability. This is because
many of the arguments for high-level content proceed by arguing that
learning can affect which visual experience one has. The idea is that the
cognitively unadulterated visual system will have limited representa-
tional powers – no doubt powers to represent only that which is likely
to be encountered by most humans in different locations and at differ-
ent times – powers that are likely to be useful to all. These representa-
tional powers will extend to properties such as shape, colour and size.
But, depending on the specifics of the environment and the skills that it
will be useful for particular humans to have, the visual system can
come to represent new objects and properties via repeated exposure to
them and via top-down influences from the cognitive system.
7
Some people think that states of the cognitive system, such as beliefs and desires
can have phenomenal character; others deny it. Those that deny it that wish to
defend the idea that experiences only have low-level content would have to deny
that there was any change in phenomenal character in such cases.
32 FIONA MACPHERSON

For example, plausibly, the visual system does not come ready-made to
represent pine trees or the specific individual that is my brother. This is
a type of tree and a particular human that not all humans will encoun-
ter. If all such specific representational abilities had to be built-in to
the visual system it would be enormous and unwieldy. But, perhaps on
repeated exposure to pine trees or to my brother, a subject can come
to notice features that all and only such trees have and that all and
only that person has. And perhaps the subject’s knowledge of these fea-
tures can feed into their visual system so that they come to have visual
experiences that are sensitive to those features, at least sensitive in a
way that they were not before. In this way, the subject would come to
have visual experiences that they did not have before – visual experi-
ences that represent pine trees or my brother. In other words, a very
plausible mechanism for visual experiences coming to have high-level
content is that the visual system is penetrated by the cognitive system.
Of course someone might be a high-level theorist and hold that the
visual system comes ready-made to make pronouncements about all
sorts of contents, including those involved in high-level content. Or
they may think that although the visual system can be altered during
one’s lifetime, it is not altered by cognition. (Perhaps they think it is
altered by some non-cognitive process, such as an associative process
that takes place solely within the visual system.) Nonetheless, high-level
content theorists are likely (though not compelled) to want it to be the
case that cognitive penetration of perceptual experience can occur.
Those who believe that visual experience represents only low-level
contents have one less reason than those who believe visual experiences
can represent high-level content to think that experience is cognitively
penetrable. For if high-level contents cannot feature in visual experi-
ence then the postulation of penetration is not required to explain
alleged high-level contents in visual experience – instead, such content
is to be denied. However, such a theorist needn’t be committed to
cognitive impenetrability. Perhaps they think that cognition can affect
the representation of various low-level contents, such as shape.
For example, there is some evidence that whether one is subject to the
Mu
¨ller-Lyer illusion depends on the nature of the environment that
one has grown-up in. For example, it seems that if one has not grown-
up in a ‘‘carpentered’’ environment (one that contains many right
angles) then one is less likely to be subject to this illusion.
8
This
8
See McCauley and Henrich (2006). The evidence is not to my mind conclusive but
further discussion of this case takes me too far from my aims in this paper. None-
theless, I will return to consider whether the mechanism for cognitive penetration
that I postulate could account for this case if it is a case of cognitive penetration in
section 5.
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 33

evidence has been taken to suggest that people who live in carpentered
environments have implicitly learned certain facts about typical rela-
tionships between distances, lengths, and the appearance of angles
where three lines or surfaces appear to converge, and thus beliefs with
contents consisting in these facts have become part of their cognitive
system. These facts, it is postulated, feedback down and influence the
lengths that our experiences represent certain lines as having – some-
times inappropriately, as when the Mu
¨ller-Lyer illusion is experienced.
Nonetheless, candidate cases of low-level cognitive penetration have,
in the past, been thought difficult to find – far more difficult than high-
level ones. Thus, there has often been little pressure for someone who
believes only in low-level content to think that cognitive penetration
occurs. But, as will become apparent, the case that I think is a persuasive
case of cognitive penetration does involve low-level content – content
about colour. Thus, I think low-level theorists who do not believe in
cognitive penetration need to re-consider their position.
3. Responses to Alleged Cases of Cognitive Penetration
In this section, I outline some cases that have been thought to challenge
the idea that perception is cognitively impenetrable. I do so in order to
show typical ways that a defender of cognitive impenetrability can defend
their thesis against such attack. I do not intend to cover all cases of
alleged penetration. In omitting some cases I do not mean to suggest that
they are not powerful challenges or ones that do not require addressing
by a defender of cognitive impenetrability. My goal here is only to eluci-
date the common ways in which many sorts of potential counterexamples
can be defused. These ways of defusing alleged counterexamples are ones
that one might think one can use to respond to the case about colour that
I will consider in more detail in the next section. But, as we will see in that
section, these ways of replying are not good ways of replying to the col-
our case – which, together with the fact that colour is a low-level prop-
erty, is what makes it such a powerful example.
A common form of alleged counterexample to the cognitive impene-
trability thesis consists of two experiences that are had by a subject at
different times that intuitively have different phenomenal characters.
Then a case is made that the best explanation for this difference is that
the cognitive system of the subject is different at those different times.
For example, in the previous section, I discussed the view that there
can be high-level content in perceptual experience. Some arguments
for the existence of high-level content in experience can be elucidated
in a way that makes them arguments that perceptual experience is
cognitively penetrated.
34 FIONA MACPHERSON

Consider, for illustrative purposes, the following argument based on
Siegel’s (2005) argument concerning pine trees. Imagine a scenario in
which someone gains a recognitional capacity – the capacity to tell that
a tree is a pine tree by looking at it. Consider the experience of the
subject before they have the recognitional capacity when they look at a
pine tree E
1
, and call the experience they have after gaining the
capacity when looking at the same tree from the same position, in the
same lighting conditions, when their eyes are working similarly, and
when the same parts of the scene are being attended to, E
2
. It is plausi-
ble that the phenomenology of the subject will be different overall
when having E
1
and E
2
. After all, when having E
1
the subject will not
know what kind of tree they are looking at. When having E
2
they will.
One explanation of this difference is that the visual experiences E
1
and
E
2
are different, and different only because the second experience repre-
sents that a pine tree is present, while the first does not. If this were
the case then there would be reason to think that gaining the recogni-
tional capacity to identify pine trees, which plausibly involves one’s
cognitive system gaining beliefs about how pine trees look and a recog-
nitional capacity to identify pine trees, affects the experience one has.
E
2
represents a pine tree, while E
1
does not and this difference in
content is manifested in the different phenomenal characters of the
experiences. Hence, cognitive penetration has taken place.
Consider another kind of case, which also conforms to the com-
mon form of potential counterexample. Certain figures are ambigu-
ous. When one looks at these figures there are two distinct ways of
seeing them. It is plausible to think that one has two different expe-
riences corresponding to these two different ways of seeing. Exam-
ples include the Necker cube (in which the line AB in the diagram
can be seen either as being on the front face of the cube or on the
back face) , the duck ⁄rabbit picture and the dol phin ⁄naked couple
picture.
One might think that the two different experiences had when seeing
the figure in the two ways possible come about because of influence
from the states of one’s cognitive system. For example, with practise,
one can learn to have one of the two different experiences almost at
will – one can voluntarily change how the figure looks. It could be
argued that here one’s thoughts about which experience one wants to
have – or one’s thoughts about what one wants to experience –
changes one’s experience. Another way in which one might think that
one’s experience is influenced by one’s cognitive system concerns
which experience one has when looking at the figures, or which expe-
rience one is most likely to have or has first. For example, it is some-
times claimed that adults typically have the experience of the naked
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 35

couple when they first look at the dolphin ⁄naked couple ambiguous
figure, whereas children typically have an experience as of the dol-
phins. If this is true, one might think that it is explained well by the
fact that children have very different thoughts, concerns, expectations
and proclivities compared to adults.
There are two main ways to respond to these types of alleged coun-
terexamples to the thesis that perceptual experience is cognitively
impenetrable. The first way is to deny that the two perceptual experi-
ences in question have different phenomenal characters. One might
claim that instead there is only a difference in the judgments one makes
about the world or the judgments one makes about one’s perceptual
experiences. (Then it is open to one to either claim or deny that these
judgmental differences bring with them some form of phenomenal
differences). For example, in the pine tree case one might claim that
the difference before and after one gains the recognitional capacity is
not that E
1
and E
2
are different but that one’s judgments are different.
One judges that there is a pine tree in one case and one doesn’t judge
that in the other case. Of course one might make additional judgments
36 FIONA MACPHERSON

in the case one has the recognitional capacity for identifying pine trees.
For example, one might make subtle judgments about the outline shape
of the tree and its leaves or their colour.
In the ambiguous figures cases, one might hold that there are not
two experiences associated with seeing each of the figures. Rather, there
are two different judgments that people have when looking at the
figures – for example, that there is a picture of a duck present or that
there is a picture of a rabbit present. On account of making these
different judgments people misjudge that their experience is changing.
Perhaps certain features of the experience become more salient to the
people on one occasion compared to another and so the people judge
that the picture is of different things. Perhaps their doing this helps to
explain why people think that their experience itself is changing when
in fact it remains unchanged.
The second way to respond is to agree that there is a difference
between the phenomenal characters of the perceptual experiences but
to hold that this is not best explained by cognitive penetration. For
example, in the case of the pine trees, one might hold that in gaining
a recognitional capacity for identifying pine trees one learns where to
focus one’s attention – perhaps on the outline shape of the tree or
leaves, etc. – in order to identify the tree. As a result of this change
in focus of spatial attention, the phenomenal character of one’s expe-
rience is altered. As discussed in the previous section, however,
changes of spatial attention, even if these are brought about by a
change in one’s cognitive system, are typically not counted as cases of
cognitive penetration. Similarly, in the case of ambiguous figures, one
might think that one’s judgment about what the figure is a picture of
affects one’s eye movements or where one focuses attention on the
figure. Again, this would result in a difference of experience – but it
is a case where one’s judgment may affect one’s experience that is not
a case of cognitive penetration. An alternative explanation in the case
of ambiguous figures is that the early visual system can autonomously
process some inputs in two distinct ways yielding two distinct experi-
ences, but that these different ways of processing are unaffected by
the cognitive system.
Which method of explaining away a potential counterexample is
best will depend on the details of the case. I don’t wish to defend
particular responses to the examples above save to note, without
argument, the fact that I find the first strategy more plausible in the
pine tree case and the second strategy more plausible in the ambigu-
ous figure case. However, there is one case that I believe is not well
explained by these types of response. It is to this case that I turn in
the next section.
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 37

4. The Colour Counterexample
In 1965, Delk and Fillenbaum published a study that is a candidate for
being a case where subjects’ beliefs about the characteristic colour of
an object can affect the colour that they experience that object as
having.
9
Their experiment is, somewhat surprisingly, not discussed as
often as it should be in the literature on cognitive penetration, given
how persuasive an example of it I believe it to be. In this section,
I relate the experiment and explain why the standard strategies outlined
in the previous section, designed to explain away alleged counterexam-
ples, do not seem to explain away this case.
The experiment was carried out as follows. The experimenters took
a sheet of paper of a uniform orange colour. They cut out shapes of
various objects from it. Some of these objects were characteristically
red: a heart (a love-heart shape), a pair of lips, an apple. Some were
not characteristically red: an oval, a circle, a square, an ellipse, a
horse’s head, a bell, a mushroom. One at a time, the cutout shapes
were placed in front of a coloured background that could be altered.
The background could be changed from a yellow colour through
orange to a red colour by twisting a nob. The subjects were told to
instruct the experimenter to make the background colour more yellow
or more red until it was the same colour as the cutout shape in front
of it, so that the cutout shape could no longer be distinguished from
the background.
10
When the cutout shape of an object that had a
characteristically red colour was placed in front of the background, the
subjects selected a background colour that was more red than the
colour that they selected the background to be when the cutout shape
was of an object that didn’t have a characteristically red colour. When
the cutout shape was of an object that didn’t have a characteristically
red colour the subjects’ instructions yielded a background that was less
red and more yellow. The experimenters claimed that:
One is led to the conclusion that past association of color and form
does in some way influence perceived color, since that is the one
9
This study was based on similar but less conclusive studies carried out along the
same lines. See Duncker (1930) and Bruner et al. (1951). Variants of this experi-
ment have been carried out more recently achieving similar results. See Hansen et
al. (2006).
10
In fact the subject had to look at the cutout against the coloured background
through a sheet of wax paper that was almost, but not quite, transparent. The pur-
pose of the wax paper was to slightly blur the boundary area of the cutout shape
against the background. I believe this detail is unimportant for my discussion of
the case. One can imagine the experiment being repeated in modern conditions on
a computer screen where such paper would not be required.
38 FIONA MACPHERSON

respect in which the figures did clearly differ. (Delk and Fillenbaum
1965: 293)
Although Delk and Fillenbaum don’t themselves suggest a mechanism
that explains why this happens, one might think that what is happen-
ing is that the subjects’ beliefs, that certain of the cutout shapes were
shapes of objects that were characteristically red, penetrated their
perceptual experience of those cutout shapes thereby altering the
content and phenomenal character of those experiences. The subjects
had experiences of those cutout shapes that represented them as being
more red than they really were. This did not happen when the cutout
shape was not of a characteristically red object. In that case, the sub-
jects had accurate experiences of the orange colour of the cutout
shape.
Should we accept that this is a case of cognitive penetration of
colour experience? The case is rather persuasive for neither of the two
strategies discussed in the section above that one might use to under-
mine the idea that cognitive penetration is occurring is a very attractive
alternative explanation of what is going on in this case.
The first strategy one could employ is to claim that there is not a
difference in the subjects’ experiences of the characteristically red and
not characteristically red objects; there is simply a change in their judg-
ments. Thus, one could claim that the experiences of the subjects when
they look at the cutout shape of a characteristically red object are not
as of a more red colour than the cutout shape really is and are not
different to their experiences of the colour of the cutout shape of
objects that are not characteristically red. Instead, one might claim that
the subjects’ experiences represent all the cutout shapes accurately as
the shade of orange that they are. The subjects merely judge that a
cutout shape is more red than it really is when it is a cutout of a
characteristically red shape.
However, this explanation requires the postulation of a misjudg-
ment on the part of a subject. The subject misjudges that the colour
of the orange cutout and the colour of the red background it is on
are the same colour when, according to this strategy, the subject’s
experience doesn’t represent them as being the same. This type of
misjudgment is not one of the sort that one commonly gets in illu-
sions – where one judges the world to be one way when it is not
because one’s experience misleads one by being inaccurate. Rather, in
this case the misjudgment in question involves judging that the world
is a certain way (the shape and its background are the same colour)
when not only is it not that way but also one’s experience tells one
that it is not that way. One’s experience tells one – represents – that
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 39

the colours are different, according to this strategy. Thus, subjects are
not only misjudging the way the world is, they are doing so in the
face of their experience telling them accurately how the world really
is. Thus, one might say, they are misjudging the nature of their expe-
rience too.
Note that to employ this type of strategy in the pine tree case,
discussed in the section above, one didn’t have to attribute any mis-
judgment or error to the subject – simply an accurate judgment that
a pine tree was present after the recognitional capacity was gained,
and a lack of judgment beforehand. In addition, the type of mis-
judgment that this response to the colour case requires is not an
explicable one – as was the case when this strategy was employed to
explain the ambiguous figures example, also discussed in the section
above. (In the ambiguous figures case, the subject’s judgment that
his or her experiences are different was explained by the fact that he
or she judges that the picture is of different things on different occa-
sions.) Rather, the type of misjudgment that the experimental set-up
requires one to attribute to the subject is brute and inexplicable and
of a very fundamental kind. To see this, consider the types of error
a subject might make on the basis of having an accurate colour
experience.
A subject might have an accurate experience of an orange colour
and judge that the colour is orange, but due to mere incorrect word
selection say that the colour is ‘‘red.’’ Such cases occur and do not
involve an error of judgment only an error of expressing one’s judg-
ment. This is not the type of error that must be postulated to explain
away the alleged case of cognitive penetration under consideration.
Second, a subject might have an accurate experience of an orange
colour but miscategorise that shade as being a shade of red. Nonethe-
less, the subject might alter a background colour to be a shade that
accurately matches the miscategorised shade of colour. This type of
misjudgment seems perfectly possible. It would appear to be one in
which, in one sense, the subject is aware of exactly the nature of their
colour experience, which they demonstrate via the colour-matching
task. Their error is simply one of which colour category to assign the
shade of colour to – red and not orange. This is also not the type of
misjudgment that must be posited by the person who wants to explain
away the apparent cognitive penetration going on in the colour case
using this strategy.
Rather, the type of misjudgment that must be going on is one in
which a subject has an accurate experience of an orange colour (of the
cutout) and an accurate experience of a shade of red (of the back-
ground), nonetheless, when looking at both the shades simultaneously,
40 FIONA MACPHERSON

they judge that the shade of red and the shade of orange are the same
shade.
11
If the subject were misperceiving one of the colours, the
misjudgment would be explicable, but this explanation is not available
as the case is one in which, by hypothesis, the subject’s experiences are
completely accurate. What we are to imagine is that the subject’s expe-
riences of the colours are different, yet the subject judges the colours to
be the same. While such cases do not seem impossible, they involve
attributing a gross form of misjudgment to the subject. This is not a
mere miscategorisation of which colour a shade is. The subject is mak-
ing a large and brute error of judgment despite their experience. We
have no explanation of why the subject makes such an error – or why
they make such errors in the systematic way the experiment shows they
do. One might think that an explanation lies in the fact that the sub-
jects believe that the cutout is of a shape that is characteristically red
and that this belief affects their judgment but, firstly, that doesn’t
explain why they should make this gross error when their experience is
telling them otherwise – for frequently our judgments are corrected by
our experiences. Secondly, this alleged explanation is the very hypothe-
sis under consideration, namely that the subject misjudges despite their
experience, so it is hardly a suitable candidate for explaining why the
subject so misjudges that when, at the same time, their experience is
telling them otherwise. That the subject does make this gross error
when they believe the cutout’s shape is of a characteristically red object
is still a fact that stands in need of explanation. Because of this, and
because there is no independent evidence that the subject’s experiences
of the cutout and the background it is on are different, the postulation
of this error and the postulation of the difference in experience with
respect to the shape and the background it is on looks to be ad hoc – a
postulation only to save the theory that perceptual experiences are cog-
nitively impenetrable.
Compare the above explanation to the alternative – that cognitive
penetration is taking place. If cognitive penetration is taking place the
subject merely misperceives one of the colours (the cutout of the shape
of a characteristically red object) and so has experiences as of the same
colour when looking at that cutout and the background. This case
involves no postulation of error in the subject’s judgment. Rather it
11
The only alternative would be to suppose that the subject misperceives the back-
ground. But if that were the case then that experience would appear to be being
cognitively penetrated by the belief about the characteristic colour of the heart-
shaped cutout. Not only would that be odd, it would be another alleged case of
cognitive penetration that would have to be explained away. Explaining it away
would presumably be done by one or other of the strategies that I am outlining
here, but these strategies would face exactly the same problems as the ones I am
outlining here for them.
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 41

merely involves an inaccuracy that the subject can’t be held responsible
for – an inaccuracy in their experience. On this account, the subject
accurately reports the way the world looks to them – even if the world
is not that way.
Thus, explaining away the alleged cognitive penetration of experi-
ence by means of this strategy involves, in this case, the postulation
of a gross, brute and inexplicable error on the part of the subject –
an error on the personal level that we think subjects are responsible
for. No doubt many people think errors of this kind are possible, but
this view predicts that these errors occur in a systematic way – for
example they will always occur in the conditions that the experiment
specifies. And that thought is unpalatable, for what then ensures any
of our judgments reflect our experience? The alternative explanation
involves no such postulation of brute, inexplicable error on the part
of the subject. For this reason, the explanation that tries to explain
away cognitive penetrability is ad hoc and far less attractive than the
alternative.
Furthermore, note that colour is a low-level property – it is a
property that all people agree is represented by visual experience – as
opposed to a high-level property, like a natural kind property. There-
fore a common strategy that is employed by low-level theorists to
maintain that two experiences are the same, and that they represent the
same properties, cannot be employed here. The strategy is to claim that
any evidence that the experiences are different is really only evidence
that the contents of judgments formed on the basis of those experiences
are different, on the grounds that experiences cannot represent the
properties in question, as they are high-level properties. But this strat-
egy can’t be applied to this case for the properties at issue – colour
properties – are low-level properties.
I have just been considering the first broad strategy that might be
employed to explain away the case of colour perception that seems
to show cognitive penetration takes place – postulating that the
experiences that the subject has of the two different kinds of cutouts
are the same and that it is their judgments about the colour that
differ. The second broad strategy that one might use to undercut the
idea that the case under discussion is a case of cognitive penetration
is to accept that the experience of the cutout shapes of characteristi-
cally red objects is different from the experience of the cutout shapes
of the objects that are not characteristically red, but to claim that
this fact is not explained by cognitive penetration. There are two
factors that one might appeal to in order to explain what other
than cognitive penetration accounts for the difference between the
experiences:
42 FIONA MACPHERSON

(i) a difference in the cognitive states of the subject, but in a way
consistent with the cognitive impenetrability of perceptual
experience (by a shift in the location of one’s attention) or
or
(ii) some factor other than the differing cognitive states of the
subject.
Let us look at each in turn.
As we saw in section three above, one might claim that the subject’s
beliefs alter the focus of their attention and this is responsible for the
change in their experience. At the same time, we noted that such altera-
tions in experience by cognition by means of attention are not counted
as cases of cognitive penetration. However, I will argue that the kind
of difference in attention that will have to be postulated to successfully
explain the change in experience in the colour case is rather special and
that it is not one that is clearly compatible with the cognitive impene-
trability of perceptual experience.
What kind of change in attention that results in a change of experi-
ence is compatible with the cognitive impenetrability of perceptual
experience? It is changes in spatial attention that people have in mind.
Indeed, I know of no other discussion of changes in attention in this
context except for changes in spatial attention. One type of case that is
discussed in this regard is the case of ambiguous figures. Consider the
duck ⁄rabbit ambiguous figure. It is often held that the difference in
experience that one has in seeing the figure as a duck and then as a
rabbit is in where one focuses one’s attention. For example, someone
might hold that if one focuses attention on the left-hand side of the fig-
ure then one will see the figure as a duck, while if one focuses attention
on the right-hand-side of the figure then one will see the figure as a
rabbit. (No doubt that explanation is in fact too simplistic and patterns
of spatial attention are required to fully explain the case, but it will do
no harm to ignore such complications here.)
Would a change in where attention is focused be a good explana-
tion of why experiences of uniform patches of colour were different
on different occasions? No – for the colour is uniform all over the
cutout, so a change in attention would not plausibly yield a change
in the experience of colour. One might think, therefore, that this
strategy is doomed to failure. However, this would be too quick.
What we should conclude is that if some form of attention shift is
going to explain this case it must be some form of attention other
than spatial attention. Reflection on the case, I believe, will show that
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 43

the best explanation involving attention that one can give here is that
when one sees the cutout shapes of objects that are characteristically
red, one pays attention to the redness within the orange colour of the
paper at the expense of the yellowness within the orange colour of
the paper. When one sees the cutouts of the objects that are not
characteristically red one doesn’t pay special attention to the redness
within the orange colour. (Orange is a binary colour, which is to say
that it is seen as a combination or mixture of yellow and red.
12
Thus
orange is experienced as having a red component and a yellow com-
ponent. It is focusing attention on the red component, at the expense
of the yellow component, that I have in mind here.) This change in
attention towards the red component in the orange colour of the
paper might alter the nature of one’s experience and this would
explain well why one chose a more red colour as a match for some
of the cutout shapes compared to others.
However, while it is frequently held that if a cognitive state changes
the location that one attends to and thereby changes one’s experience
then it needn’t count as a case of cognitive penetration, it is not
obvious that other changes in attention caused by a cognitive state
should not count as cases of cognitive penetration. To see this, note
that the reasons that explain why people think that changes of spatial
attention directed by one’s cognitive states should not count as cases of
cognitive penetration are two-fold. The first is that such cases are akin
to selecting what spatial location to attend to by means of changing
the position of one’s body or eyes. But the kind of attention that has
to be postulated to explain the colour case is selection of one feature
or property of an object (its redness) that is instantiated at the very
same location as another feature or property of an object (its yellow-
ness). Thus, it is not the sort of selection of one feature over another
that could be brought about by a change in the position of one’s body
or eyes. The second reason people have for allowing a change in the
focus of spatial attention cased by a cognitive state to be consistent
with the cognitive impenetrability of experience is that they think that
allocation of attention to a spatial location occurs before perceptual
processing occurs. The thought is that the nature of the stimulus, the
nature of the perceptual conditions, the state of the sensory organ and
where attention is focused are all conditions that determine what will
get perceptually processed – but the perceptual processing itself, that
proceeds after these conditions are in place, is impenetrable. However,
it is far from clear that attention to the redness within a patch of
orange is an attentional effect that is a condition that fixes what will be
12
See Hardin (1988: 39).
44 FIONA MACPHERSON

perceptually processed in advance of any such processing taking place.
On the contrary, it is reasonable to think that attention to the redness
within the orange only occurs once perceptual processing has got
underway, when orange has been registered by the brain or, if one is
thinking of attending as something that the subject does (as opposed
to something done by a subpersonal mechanism) when the subject
becomes aware of the orange colour. After all, how can there be
attention to a feature of the orange colour if one’s brain has not
detected the orange colour or if one is not aware of – experiencing –
the orange colour? In this respect, this form of attention is rather
different from spatial attention, about which there is evidence that
one’s attention can be focused on a location prior to one’s having a
visual experience. For example, if one had one’s eyes closed (and thus
suppose one had no visual experience) a noise might lead one to attend
to the apparent location of the noise and, on subsequently opening
one’s eyes, one’s experience might in part be the way it is because one’s
attention is focused on that location. The thought is that spatial cues
in other modalities, or spatial cues in the same modality prior to
having the particular perceptual experience in question, or voluntary
thought of a location alone, can focus attention on a location
and thereby alter subsequent visual processing and, hence, subsequent
experience.
13
In short, if one adopts the strategy of claiming that cognitive
penetration is not taking place in the colour case because, although
one’s experiences of the colours of the shapes are different, this can be
explained by one’s cognitive states affecting attention, then certain
further explanatory tasks become incumbent upon one. First, one needs
to explain what kind of change in attention could change the nature of
one’s experience. Second, one has to ensure that the kind of change of
attention cited is one that is compatible with the cognitive impenetra-
bility of experience. I have given some reasons in the above paragraph
to think that the most plausible form of change of attention that would
explain one’s experience in the colour case is not a form of spatial
attention. It is attention to one of the colour components within a
binary colour – attention to the redness that exists within orange. And
I have argued that this is not a form of attention that can be easily
construed as not being a form of cognitive penetration.
Indeed, consideration of this case brings to the fore the more general
difficulty of establishing in any case – even in the case of spatial attention
– whether the contribution of attention to perception really is a
13
Evidence for this type of effect has been found, for example, in: Spence and Driver
(2004), Santangelo et al. (2006), Santangelo et al. (2009) and Sto
¨rmer et al. (2009).
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 45

pre-perceptual phenomenon. As this is a difficulty that plagues all forms
of attention then so much the worse for any strategy of explaining
away alleged cases of cognitive penetration that relies on the claim that
changes in attentional focus caused by changes in cognitive state can
occur without cognitive penetration.
14
I turn now to consider (ii), that is, whether some factor other than
the differing cognitive states of the subject could be responsible for the
difference between the experiences. Delk and Fillenbaum explicitly say
that they tried to ensure that the only relevant difference between the
cases where the subjects were presented with cutout shapes of charac-
teristically red objects and non-characteristically red objects was the
subject’s beliefs. For example, they tried to ensure that the size of all
the cutouts was similar and that each group of characteristically red
and non-characteristically red cutouts contained members that had
both curved and straight lines. I think that it is reasonable to believe
that they ensured that no factor external to the subject was responsible
for the different behaviour of the subject towards the characteristically
red and non-characteristically red cutouts, but did they ensure that
there was no relevant difference within the subject apart from the
subject’s beliefs?
One thing undetermined is whether the subjects’ visual systems
reacted differently when they saw the cutouts of the characteristically
red objects compared to the non-characteristically red objects but not
because of influence from the cognitive system. It might be that the
early visual system, autonomously from belief or other cognitive states,
alters the colour experiences of characteristically red shapes, to make
them appear more red than they really are. This might happen in
accordance with associationist principles, so that it is past exposure to
a certain shape having a certain colour that has altered the way the
visual system processes that shape’s colour.
One motivation for thinking that this might be the case comes from
considering what would happen if you suddenly lost your belief that
one of the characteristically red shapes, say the heart, was red (for
example, a God appeared to you and told you that this was the case).
My intuition says that it is plausible that you would still undergo the
colour effect. You would still experience orange hearts as more red
than orange non-characteristically red shapes.
15
If this is right, then
perhaps it is not the cognitive system that is responsible for the differ-
ence in the colour experience that subjects have.
14
The question of whether attention is a pre-perceptual process or not is discussed in
Raftopoulos (2005) and Rowlands (2005).
15
Thanks to Tim Bayne for making this motivation for the view clear to me.
46 FIONA MACPHERSON

There are two ways in which one can reply to this suggestion for
what other than cognitive penetration explains Delk and Fillenbaum’s
results. The first is to agree that perhaps beliefs with the content that
hearts are characteristically red cannot penetrate your experience.
However, one might think that other beliefs can penetrate it. Perhaps
the beliefs that can penetrate are ones that it would be very difficult
for a God to overturn (just by telling you, without tampering directly
with your brain). Examples of such beliefs are the belief that most of
the hearts I have experienced have seemed red and the belief that for
some time I believed that most hearts were red. A very similar
response would be to claim that it is not any specific belief that is
responsible for the effect on the visual system, rather, because of your
(at least apparent) exposure to red hearts, the concept ‘‘red’’ gets
primed when you have an experience as of a heart. Your cognitive
system, in this primed state, then affects the visual system. This does
not require any particular beliefs to be present in order for cognitive
penetration to occur. In fact, one way to spell out what it is for
the cognitive system to be so primed is that one is disposed to bring
to mind some beliefs or other about redness, but which ones is
irrelevant.
The second reply that one can make to the suggestion that the visual
system affects your experience of characteristically red shapes without
the cognitive system penetrating it is to admit that perhaps in some
cases this is the mechanism at work, but to claim that there are an
important class of cases where this is not a plausible explanation. This
reply is convincing because there are cases where the objects that elicit
the effects potentially explicable by cognitive penetration are ones that
have to be classified by the cognitive system in one way rather than
another for the effect to occur. They are cases where it is implausible
to think that low-level vision can do this categorisation alone and
belief, desire or other cognitive, conceptual states play a crucial role in
the occurrence of the effects.
Examples of such cases include the shapes of characteristically red
objects used in the colour case, other than hearts, such as apples. While
it may be plausible to think that the visual system is sensitive to heart
shapes and autonomously responds to that shape making it seem more
red that it is, it is far less plausible to think that the visual system can
classify certain objects as being apples and autonomously alter the
apparent colour of such objects. The reason is that classifying some-
thing as an apple doesn’t depend just on simple features, such as shape,
that the visual system is responsive to, for many things have the shape
of apples, such as cherries and other fruits. And while the visual system
can respond to shape, it surely can’t classify an object as an apple, as it
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 47

is not sensitive to the features that are required to do so. It is surely
only the cognitive system that is capable of doing such classificatory
work.
An even better example of objects that elicit the effects potentially
explicable by cognitive penetration, and are ones that have to be clas-
sified by the cognitive system in one way rather than another for the
effect to occur, are certain faces. A series of experiments carried out
by Levin & Banaji (2006) showed that greyscale pictures of faces that
have stereotypical features of white people and similar pictures of faces
that have stereotypical features of black people are matched by sub-
jects for lightness to different samples of grey, even though the pictures
of the faces have the same surface luminance. Subjects were instructed
to change a patch of grey that could be altered from dark to light to
match a picture of a face placed beside it. The faces that looked like
faces of white people were matched to lighter shades of grey than
those of black people. Remarkably, the effect appears to increase when
one knows about the effect and stares hard at the faces, as readers will
be able to verify for themselves by looking at the examples of faces
used in the experiment re-published below. In a variant of the experi-
ment, a racially ambiguous face was labelled either as the face of a
white person or the face of a black person and this factor alone
determined what shade of grey the subjects chose as a match for the
lightness of the face. This second result is particularly significant, as
the categorisation was clearly done at the cognitive level as it was the
labelling of the face that was responsible for the effect. Levin and
Banaji state:
White faces were consistently judged to be relatively lighter than Black
faces, even for racially ambiguous faces that were disambiguated by
labels. Accordingly, relatively abstract expectations about the relative
reflectance of objects can affect their perceived lightness. (2006: 501)
These face cases provide examples involving colour matching, as did
Delk and Fillenbaum’s experiment, which we have already seen lend
strong support for the thought that two different experiences of colour
(in this case lightness) are being had by subjects. At the same time they
provide an excellent example where the categorisation of the object,
which is necessary for the different experiences of lightness to be had,
takes place at the cognitive level.
Three faces used and published in Leven and Banaji (2006). The ste-
reotypically black face on the left has the same luminance as the ste-
reotypically white face in the middle. The face on the right is an
example of a racially ambiguous face.
48 FIONA MACPHERSON

Thus, experiments of the type carried out by Delk and Fillenbaum, and
Levin and Banaji, can provide very good reasons for thinking that colour
experience is cognitively penetrable. The typical ways in which one might
explain away these cases face clear difficulties. So, I believe that the sim-
plest and best explanation of these cases is that cognitive penetration is
occurring. A subject’s experiences of the colour of different objects is not
the same, despite the colour of the objects actually being the same. This
is because their experiences are penetrated by either the subject’s belief
that some of the objects are of a type that have a characteristic colour, or
by some state of the subject’s cognitive system in which it is primed to
make certain beliefs about that characteristic colour conscious.
5. A Mechanism for Penetration
If one is persuaded that the colour case outlined in the previous section
is a case of cognitive penetration, then the question arises of how a
state of the cognitive system can affect perceptual experience. It
certainly doesn’t seem that we can affect our perceptual experiences at
will. If only we could not experience the sound of the fingernails scrap-
ing down the blackboard, or taste the horrible mushrooms hidden in
the stew, or see the litter spoiling the otherwise beautiful view, just by
wishing! Similarly, how we expect the world to be is often not how we
perceive it to be, either to our delight or disappointment. Resistance to
the thought that cognitive penetration can occur is, I think, in part a
concern about the nature of the mechanism of penetration. In this
section, therefore, I propose a plausible psychological mechanism
to explain cognitive penetration. It is a two-step process and I will
bring forward evidence, independent of the phenomenon of cognitive
penetration, that shows that each step can occur. In virtue of the mech-
anism having these two steps, I call it indirect. I think that the mecha-
nism may make the idea that cognitive penetration occurs more
palatable to those who are (or were prior to exposure to the colour
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 49

case of section 4) apt to think that cognitive penetration does not
occur. And for those people who think cognitive penetration is wide-
spread, this mechanism may explain a large number (although I don’t
claim all) of the cases in which they think cognitive penetration occurs.
The first step of the mechanism involves our cognitive states causing
some non-perceptual state with phenomenal character to come into
existence or to alter the phenomenal character of some existing non-
perceptual state that has phenomenal character.
16
(By ‘‘non-perceptual
state’’ I mean here a state not involved in perceiving the world. On
this disambiguation of the term, hallucinatory experiences are non-
perceptual states. Of course the term ‘‘non-perceptual state’’ often is
used otherwise, namely, to refer to states other than the experiences
involved in both seeing and hallucinating. But it is convenient for my
purposes here to use the term in the more inclusive sense.) Such cases
are commonplace. Here are some examples:
•You are imagining nothing. I then ask you to imagine looking
at George Square in Glasgow from Queen Street station.
You comply. Your cognitive state has caused a non-perceptual
state with phenomenal character to come into existence. You
deliberately imagined what you did.
•You are daydreaming about sailing down the river Clyde. You
are imagining the sights and sounds and the feeling of smirr on
your face. Suddenly you think of me and find yourself imaging
sailing down the Clyde with me. Your thought about me has
changed your daydream, and altered its phenomenal character.
This whole imaginative experience might not have been deliber-
ate. The imaginative experiences might have come unbidden.
•You dream of me because you met me today for the first time.
•You visually hallucinate me and part of the explanation is that
you know me well.
It is widely agreed that imagining, dreaming and hallucinating always
or frequently involve being in a state with phenomenal character.
Philosophers are wont to imagine that some such states have the very
same phenomenal character as some perceptual experiences. But often
16
As we will see later, it might be best to think that what happens in the first step is
that our cognitive states cause the initiation of an imaginative process that would
normally result in an imaginative experience with phenomenal character.
50 FIONA MACPHERSON

these states are very unlike the typically perceptual experiences that we
have. For example, the phenomenal character of imaginative experience
is said to be ‘‘weaker’’ or ‘‘less lively’’ than that of perceptual experi-
ence. And imagination typically does not have the force of presenting
the world as being the way imagination represents it to be. In addition,
imagination can be much less determinate than typical perceptual
experience. Hallucinations are sometimes recognised as such for their
bizarre content and, as J. L Austin pointed out, dreaming that one
meets the Pope is rarely, if ever, like really meeting the Pope.
17
Imagination, dreams and hallucination clearly frequently involve
some phenomenal character, and there is no doubting that the content
and phenomenal character of these states is often generated, affected by,
and dependent on our beliefs and desires. In this sense, these states are
cognitively penetrable. The content of such states is overwhelmingly con-
tent dependent on one’s cognitive states, although there may be some
content of such states that is not so dependent. One oft cited example of
the latter is when the content of a dream is affected by one’s alarm clock
going off and the sound of the alarm may feature as an element of the
dream.
18
Another case might be one in which an evil scientist tampers
with your brain and the content of your resulting hallucination is attrib-
utable to that tampering and not to your cognitive system. But such
examples don’t detract from the thought that often the content of such
states is generated by or dependent on one’s beliefs, thoughts, desires,
expectations and previous experiences with the world. Indeed, I believe
that the content of the majority of such states are, as a contingent matter
of fact, affected by the states of one’s cognitive system.
The second step involves the phenomenal character of these non-per-
ceptual states interacting with and affecting the phenomenal character
and content of perceptual experiences. For example, in the colour case
discussed in the section above, perhaps subjects imagine the cutout
shape being red and the phenomenal character of this imaginative state
interacts with the phenomenal character of their visual experience. In
this case, it doesn’t seem plausible to suggest subjects are aware of two
states or two phenomenal characters. So the best way to put the idea is
that the contribution of the imagination and the contribution of vision
combine producing one phenomenal state. If the imagination would
have produced a state with the phenomenal character of an experience
17
See Austin (1962).
18
One might claim that the sound of the alarm isn’t part of the content of the dream
and is part of the content of one’s auditory experience. But even if this example
doesn’t provide a good example, there are others and it doesn’t detract from the
main thrust of my argument which is simply that much of the content of dreams,
hallucinations and imaginative states is determined by one’s cognitive system.
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 51

had when imagining a red object if no contribution from the visual
state had been present, and if the visual system would have produced a
state with the phenomenal character of a visual experience had when
veridically seeing an orange object if no contribution from the imagina-
tive state had been present, then the resulting phenomenal character of
the combination of vision and imagination would plausibly be the
phenomenal character of an experience had when veridically seeing a
reddish-orange object. This would explain why the subjects matched a
more red colour to cutouts of an orange shape with a characteristically
red colour compared to those that weren’t characteristically red.
The next question to address is why should one think that percep-
tion can interact with other mental processes – processes such as imagi-
nation, dreaming and hallucinating that are affected by states of the
cognitive system – to yield a state whose phenomenal character must
be explained by the contribution of both perception and those other
process? After all, one shouldn’t think that the other process provides,
say, some phenomenal redness that mixes with, say, the phenomenal
orangeness provided by vision in the way that paints can be mixed.
This is true, but shows only that this suggestion should not be the
model of such interaction. Nonetheless, we have reason to believe that
this kind of interaction between such states occurs. There are examples,
completely separate and independent of potential cases of cognitive
penetration, where we believe that the phenomenal character of a state
is brought about by contributions from both perception and either
imagination, dreaming or hallucinating.
The first example is illustrated in the Perky effect. Perky conducted
an experiment in which subjects were asked to fixate at a point on a
blank white screen and to visually imagine various objects on it, such
as a tomato, a banana, a book, and a leaf. While the subject did this,
a faint image of the particular object was projected onto the screen.
The image was above the typical threshold for visibility. The subjects
didn’t realise that this was going to happen. Nor did they report that
they saw anything at the time or when explicitly asked afterwards
whether they had seen anything or imagined everything. However,
when they reported what they were imagining, it was clear that what
was projected onto the screen affected their experience. For example,
one subject reported their surprise when they tried to imagine a
banana in a horizontal position and claimed that they had ended up
imagining it in a vertical position. Another subject was surprised as
she said that she had intended to imagine an elm leaf but ended up
imagining a maple leaf.
Perky’s experiments are typically reported as being ones in which
perceptual experiences were mistaken for imaginative mental states and
52 FIONA MACPHERSON

hence that the phenomenal character of some visual experiences are the
same as that of some imaginative experiences or at least that any differ-
ences are a difference of degree and not of kind.
19
However, others
have thought that this conclusion is too strong. Segal, who followed up
Perky’s work with subsequent similar experiments, held that what was
happening was that imagining raised subjects’ perceptual threshold so
that they did not have a perceptual experience of the image, yet uncon-
scious perception of it influenced what they did imagine.
20
What does seem clear is that the phenomenal character of the result-
ing state – be it an imaginative one, or a perceptual one mistaken for
an imaginative one, or some combination of perceptual or imaginative
state – has aspects contributed by perceptual and imaginative processes.
For example, as we have seen, the image projected on the screen
affected the reported phenomenology of the subsequent experience.
But imagination did too. One subject reported visually imagining red
vein-like markings on the leaf, when none were present on the image
projected on the screen. Another subject claimed that the book they
were imagining had text on it, when the projected image of the book
had no writing on it. Colours from the projected image can affect
the resulting experience. Segal (1972, as reported in Thomas (2008))
performed an experiment like Perky’s in which she asked subjects to
imagine the skyline of New York. An image of a tomato was projected
onto the screen. Observers didn’t report the tomato image but several
reported imagining seeing New York at sunset. Thus, we have exam-
ples where perceptual elements and imaginative elements combine to
produce the phenomenal character of what seems to the subject to be
one state – be it really a perceptual experience, really an imaginative
experience, or some amalgam of the two.
The second case has already been discussed briefly – that of what is
often described as the incorporation of external stimuli into dreams. As
with the Perky case, I am not interested in whether it is correct to say
that a subject who has, say, the sound of their alarm clock incorpo-
rated into their dream, is dreaming the sound or hearing the sound, or
some amalgam of the two. But, like the Perky case, it does not seem as
if the subject is aware of two separate states – a dreaming one and a
perceptual one. And the resulting phenomenology, of what is often
reported to be the one dream state, is made up of elements, such as the
noise of the alarm clock, that have been perceived and dream elements
which are, as discussed above, typically highly dependent on cognitive
states. For example, one can imagine that the noise of one’s alarm
19
See, for example, Thomas (2008).
20
See Segal (1971) and Segal and Fusella (1971) cited in Thomas (2008).
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 53

clock is experienced as the sound of a bin-lorry reversing in one’s
dream. Or one could imagine that a loud crash sound heard whilst
asleep might be experienced as a slightly different sound – that of
thunder – in one’s dream.
The third case is the case of hallucinations affecting visual experi-
ence. Many hallucinations are such that the objects that the subjects
hallucinate appear to be in the world, at least of part of which the
subject really is perceiving. This is true in cases in which the subjects
realise that they are hallucinating and in cases in which they don’t. For
example, in Lilliputian hallucinations, very small people are halluci-
nated and seem to be seen, for example, on the carpet or peeking out
from behind the curtains or on the hallucinator’s food.
21
Some halluci-
nations in Parkinson’s disease involve the apparent perception of
objects such as cats moving around the room between chair legs and
even sitting on the subject’s lap.
22
Real perceived objects can occlude
and be occluded by hallucinated objects. If these hallucinations are in
part caused by states of the subject’s cognitive system then, together
with perception of the environment, they seem to yield a mixed percep-
tual and hallucinatory state, the phenomenology of which is a mixture
of both perceptual and hallucinatory elements.
23
These three examples, the Perky effect, incorporation of perceived
elements into dreaming, and mixed perceptual and hallucinatory states,
suggest that the phenomenology of perceptual states can be altered by
non-perceptual states or processes or vice versa. This shows that it is
not implausible to think that processes affected by cognition can affect
the phenomenology and content of one’s perceptual experience. This
shows that the second step of the proposed indirect method for cogni-
tive penetration could plausibly occur.
So far in this section, I have proposed a two-step, indirect, mecha-
nism for how cognitive penetration may occur. First, one’s cognitive
states generate an imaginative phenomenal state or process that typi-
cally would produce such a phenomenal state. Second, the imaginative
21
See Chand and Murthy (2007).
22
See Manford and Anderman (1998).
23
Indeed, it is not obvious that these cases that involve a mixture of hallucination
and perception are not themselves cases of perceptual experiences being cognitively
penetrated. To my knowledge these cases are never discussed in the cognitive pene-
tration literature. One might resist the thought that these cases are instances of cog-
nitive penetration by holding that the non-hallucinated elements of the experience
are not cognitively penetrated. But it is not obvious to me that we should not just
think of these cases as cases of cognitive penetration heretofore not classified or
recognised as such.
54 FIONA MACPHERSON

state or process interacts with the perceptual state or process yielding
one state that is an amalgam of the two states and has a phenomenol-
ogy that has elements contributed from both perception and imagina-
tion. Thus, for example, in the colour case discussed above, the first
step would be that a subject’s knowledge that the cutout was of a
shape that is of a characteristically red object would affect the subject
causing them to imagine a red object, or generate the process that
would typically produce an imaginative experience of a red object.
The second step involves the phenomenal character of this imaginative
state, or the imaginative process that typically would produce it, inter-
acting with the phenomenal character of the visual experience of the
orange cutout shape, or the perceptual process which would typically
yield such a visual experience. The result is an experience as of a
reddish-orange colour. In this case, it doesn’t seem plausible to sug-
gest subjects are at any stage aware of two states or two phenomenal
characters. So, in this case and perhaps almost all, the relevant imagi-
native and perceptual processes simply produce one state with phe-
nomenal character whose nature has contributions from both the
imaginative and perceptual processes. If the imaginative process would
have typically produced an imaginative experience of red if unaffected
by the perceptual state, and if the perceptual system would have typi-
cally produced a visual experience as of orange if unaffected by the
imaginative state, then it is unsurprising that the phenomenal charac-
ter of the state produced by both processes would be the phenomenal
character of an experience had when accurately seeing a reddish-
orange object. Having this experience would explain why the subjects
matched a more red colour to cutouts of shapes with a characteristi-
cally red colour compared to cutouts of shapes that weren’t character-
istically red.
I would like to stress at this point that imagining may be a process
that is deliberate and initiated by an agent or it may not be. One
may choose to imagine and what to imagine, or one may simply find
oneself imagining, and one may have no choice about whether or
what one imagines. I think that the sort of imaginative process that
will typically take place in my two-step mechanism is the non-deliber-
ate, unbidden kind. Indeed, typically, one will not even realise that
one is undergoing an imaginative process or that one’s imagination is
influencing one’s experience. (The precedents for the existence of such
cases, where subjects are unaware of the combination of processes
contributing to their experience is, as mentioned above, the Perky
case, perceptual insertion into dreams, and cases of mixed perceptual
and hallucinatory states, where people don’t realise that they are hal-
lucinating.)
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 55

This two-step mechanism may explain other cases of cognitive pen-
etration if, indeed, there are others. (But perhaps it will not explain
all cases of cognitive penetration that exist – I do not argue that it
will in this paper.) For example, if we have two different experiences
when looking at pictures of ambiguous figures and this is due to cog-
nitive penetration then my two-step indirect mechanism may explain
what happens. Consider the Necker cube pictured above. Perhaps one
imagines the line AB either at the front or at the back of the cube,
or at least the process by which this imagining would take place is
set in motion. Then the imaginative experience or process interacts
with one’s perceptual process to yield one or other of the experiences
of the cube.
Another case that the two-stage model may explain is the cognitive
penetration that some people think takes place in, and explains, the
Mu
¨ller-Lyer illusion. Recall from section 2 that there is evidence that if
one has not grown-up in a ‘‘carpentered’’ environment (one that con-
tains many right angles) then one is less likely to be subject to this illu-
sion. Suppose, for the sake of argument, that cognitive penetration
does take place when one undergoes the Mu
¨ller-Lyer illusion – that is,
suppose one’s beliefs and expectations based on living in a carpentered
world, and which one would not have if one lived in a different sort of
environment, cause one to experience the lines as having different
lengths. What kind of explanation might the two-stage model give of
this case?
It may be that people who suffer the illusion are not able to help
themselves imagining lines of the upper horizontal sort in the Mu
¨ller-
Lyer figure above as being further away than the bottom horizontal
sort of line because of their experiences with corners of rooms or build-
ings. And their imagination may combine with their perception of the
two lines to yield an experience in which the top line looks longer
(because it is imagined to be further away) than the bottom line.
Another example concerns music. It is sometimes claimed that a
person who is knowledgeable about music – knows about chords,
keys, and music theory – will have a different experience of music
than someone who lacks this knowledge. Suppose that cognitive pene-
tration occurs in such a case. (I don’t argue that it does.) It would be
plausible to think that it occurs because the person imagines the struc-
ture of the notes and music corresponding to the musical theory they
know and that their imagining affects their perception of the music.
For example, musical theories are sometimes illustrated by three-
dimensional models. One example is the pitch spiral, which is a model
of Western music’s pitch structure. Ascending pitches are represented
as positions on an ascending spiral and parts of the spiral directly
56 FIONA MACPHERSON

above or below one another always correspond to notes that are one
octave apart.
If one is familiar with the model then perhaps when one hears a mel-
ody played in one octave and then in a higher octave one imagines the
same melodic structure being repeated higher-up the spiral and perhaps
this imaginative process affects one’s auditory experience.
The final example I’ll discuss comes from cricket. One might think
that a cricket umpire who is a specialist in making fine judgments
about the behaviour of the cricket ball has a different experience
of a ball hitting a batsman’s batting pad than that of a non-expert.
The non-expert may see just the ball hit the pad and then have to judge
whether the ball would have gone on to hit the wicket – something
they need to determine in order to judge whether the batsman was leg
before wicket. The expert umpire may see the ball hit the pad but also
have an experience as of the ball travelling through the space occupied
by the batsman’s leg and hitting or missing the wicket – an experience
that would be somewhat like looking at the ‘‘Hawkeye’’ animations
that are produced by computer on a screen for the third umpire who is
off the pitch and for the viewers on television. If umpires have such
experiences (and it not clear that they do, nor have I argued for it here)
then it would be reasonable to think that it was their imagination,
fuelled by their experiences of the behaviour of balls in the context of
the game and their experience of the Hawkeye pictures, in combination
Pitch S
p
iral
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 57

with the contribution from their perception, that produced this experi-
ence.
24
Alleged cases of cognitive penetration that plausibly wouldn’t be
explained by my two-stage, indirect mechanism involve expert percep-
tion where the only difference between the novice and the expert is that
the expert is accurately picking up on some objective properties that
the novice isn’t. In such cases, degrees of sensitivity to the way the
world is, rather than imagination, may provide a better explanation.
My mechanism will be better at explaining cases in which there is
misperception, illusion, mere subjective differences in perception, or the
addition of content to experience beyond that which is strictly speaking
seen (such as in the cricket umpire case).
Of course, it may not be that all imaginative states can combine with
perceptual ones to change the phenomenal character of the resulting
state. I do not suppose that occurs, in part, because I can’t change my
perceptual experiences at will by imagining. Moreover, I don’t suppose
that one can come to see any two lines of equal length as being different
in length by imaging that they unequal or by imagining that one is
further away from another. Very specific conditions for interactions
between perceptual and imaginary processes may exist. For example,
the difference between voluntary and involuntary imagination may have
some, as yet unknown, role to play, as may one’s familiarity with what
one is imagining, as may some relations between imagined and per-
ceived properties. These are matters for further philosophical reflection
and no doubt in part for psychologists to investigate.
Psychologists may, indeed, be able to test my theory of the two-stage
mechanism of indirect cognitive penetration. One way of testing it
would be to see if people in whom perception is preserved whilst visual
imagery is destroyed are as susceptible to the colour illusion and the
Mu
¨ller-Lyer illusion as other subjects in whom imaginative capacities
are preserved. There is a vigorous debate in the current literature as to
whether visual perceptual and visual imaginative processes take place
in the same part of the brain. Those that argue that they do hold that
one cannot have visual perception without visual imagination and vice
versa. However, evidence of a double dissociation between visual per-
ception and visual imagination has emerged.
25
If this evidence stands
24
Thanks to John Sutton for suggesting this example to me. (And thanks to Jona-
than Bird for setting me straight on the cricketing terminology!) In fact, this would
be a case of someone having an experience like the experiences of those with
‘‘heavenly vision’’. Heavenly vision was posited to be the vision that angels have.
They were said to be able to see solid objects as solid and yet still see other solid
objects directly behind them, without the use of mirrors, etc.
25
See Moro et al. (2008), Bartolomeo (2008), and Bartolomeo et al. (1998).
58 FIONA MACPHERSON

up to scrutiny, there would be a way of empirically disconfirming my
hypothesis: find a subject that cannot visually imagine but is still sub-
ject to experiences that we believe are cognitively penetrated and that
are so by the mechanism I propose. If the evidence dos not stand up to
scrutiny, however, then the evidence that visual perception and visual
imagination are, at least partly, processed in the same part of the brain
would help to explain why interaction between the two processes
occurs.
Another way to test my theory wouldn’t depend on finding people
who can’t do or undergo any visual imagining, rather, it would try to
determine whether people who score highly on tests that determine that
they are good at visual imagining are more susceptible to the apparent
effects of cognitive penetration, and people who score lowly are less
susceptible. For example, one could test whether people who are good
at visual imagery adjust a background in front of which have been
placed orange cut-outs of characteristically red objects to be a more
red colour than those who are bad at visual imagery. Similarly, one
could test whether there is a correlation between susceptibility to the
Mu
¨ller-Lyer illusion (among people who live in carpentered
environments) and visual imaginative abilities.
6. Conclusion
I have argued that there is one particularly difficult case to account for
by those who don’t believe that cognitive penetration of perceptual
experience can take place. The case involving colour perception cannot
be easily explained away by standard strategies for dealing with such
cases.
I proposed a two-stage, indirect mechanism that would explain
how cognitive penetration could occur. The mechanism is indirect
because the penetrating belief causes some non-perceptual phenomenal
character to come into existence, or causes some process that would
typically produce that phenomenal character to initiate. Then that
non-perceptual phenomenal character, or that process affects, or inter-
acts, with or modifies, the perceptual phenomenal character that one
has. I provided independent evidence that each step in this process
can occur.
I suspect that people who are opposed to the idea that perceptual
experience is cognitively penetrable will be less opposed to the idea
when they come to consider this mechanism because it is indirect and
proceeds by one state with phenomenal character interacting with
another or processes that each produce phenomenal character inter-
acting with one another. I also suspect that those who are generally
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 59

sympathetic to the idea of cognitive penetrability will welcome the
elucidation of this plausible mechanism. The existence of this mecha-
nism is open to empirical testing, as are the conditions under which it
operates, as I have outlined above.
26
References
Austin, J. (1962) Sense and Sensibilia, edited by G. J. Warnock,
Oxford: Oxford University Press.
Bartolomeo, P (2008) ‘‘The Neural Correlates of Visual Mental Imag-
ery: An Ongoing Debate,’’ Cortex, 44: 107–108.
Bartolomeo, P., Bachoud-Le
´vi, A. C., de Gelder, B., Denes, G., Dalla
Barba, G., Brugie
`res, P., et al. (1998) ‘‘Multiple-Domain Dissociation
Between Impaired Visual Perception and Preserved Mental Imagery
in a Patient With Bilateral Extrastriate Lesions,’’ Neuropsychologia,
36(3): 239–49.
Bruner, J. S., Postman, L. and Rodrigues, J. (1951) ‘‘Expectation and
the Perception of Color,’’ The American Journal of Psychology,
64(2): 216–227.
Chand, P. K. and Murthy, P. (2007) ‘‘Understanding a Strange Phe-
nomenon: Lilliputian Hallucinations,’’ German Journal of Psychia-
try, 10: 21–24.
Churchland, P. M. (1979) Scientific Realism and The Plasticity of Mind,
Cambridge: Cambridge University Press.
—— (1988) ‘‘Perceptual Plasticity and Theoretical Neutrality: A Reply
to Jerry Fodor,’’ Philosophy of Science, 55(2): 167–187.
Delk, J. L. and Fillenbaum, S. (1965) ‘‘Differences in Perceived Colour
as a Function of Characteristic Color,’’ The American Journal of
Psychology, 78(2): 290–93.
Duncker, K. (1939) ‘‘The Influence of Past Experience upon Perceptual
Properties,’’ The American Journal of Psychology, 52(2): 255–265.
Fodor, J. A. (1983) The Modularity of Mind, Cambridge MA: MIT
Press.
—— (1984) ‘‘Observation Reconsidered,’’ Philosophy of Science, 51(1):
23–43.
Goldstein, J. and Davidoff, J. (2008) ‘‘Categorical Perception of Ani-
mal Patterns,’’ British Journal of Psychology, 99: 229–43.
26
Thanks to David Chalmers, Parker Crutchfield and Stuart Crutchfield and audi-
ences at the University of Toronto (particularly Stephen Biggs, Mohan Matthen
and Dustin Stokes), the University of Oxford (particularly Tim Bayne), the Univer-
sity of Cornell (particularly Susanna Siegel and Nico Sillins) and at CenSes: Centre
for the Study of the Senses at the Institute of Philosophy, University of London
(particularly Ophelia Deroy) for discussing these issues with me.
60 FIONA MACPHERSON

Hansen, T., Olkkonen, M., Walter, S. and Gegenfurtner, K. R. (2006)
‘‘Memory Modulates Color Appearance,’’ Nature Neuroscience,
9(11): 1367–8.
Hardin, C. L. (1988) Color for Philosophers: Unweaving the Rainbow,
Indianapolis, Indiana: Hackett Publishing.
Levin, D. T. and Banaji, M. R. (2006) ‘‘Distortions in the Perceived
Lightness of Faces: The Role of Race Categories,’’ Journal of
Experimental Psychology: General, 135(4): 501–12.
Manford, M. and Anderman, F. (1998) ‘‘Complex Visual Hallucina-
tions: Clinical and Neurobiological Insights,’’ Brain, 121: 1819–40.
McCauley, R. N. and Henrich, J. (2006) ‘‘Susceptibility to the Mu
¨ller-
Lyer Illusion, Theory Neutral Observation, and the Diachronic
Cognitive Penetrability of the Visual Input System,’’ Philosophical
Psychology, 19: 79–101.
Moro, V., Berlucchi, G., Lerch, J., Tomaiuolo, F. and Aglioti, S. M.
(2008) ‘‘Selective Deficit of Mental Visual Imagery With Intact
Primary Visual Cortex and Visual Perception,’’ Cortex, 44: 109–
118.
Nagel (1974) ‘‘What is it Like to Be a Bat?,’’ Philosophical Review, 83:
435–50.
Perky, C.W. (1910) ‘‘An Experimental Study of Imagination,’’ American
Journal of Psychology, 21: 422–452.
Pylyshyn, Z. W. (1999) ‘‘Is Vision Continuous with Cognition? The Case
for Cognitive Impenetrability of Visual Perception,’’ Behavioral and
Brain Sciences, 22: 341–423.
Raftopoulos, A. (2001) ‘‘Is Perception Informationally Encapsulated?
The Issue of the Theory-Ladenness of Perception,’’ Cognitive
Science, 25: 423–51.
—— (2005) ‘‘Cognitive Penetrability of Perception: A New Perspective,’’
in A. Raftopoulos (ed.) Cognitive Penetrability of Perception: Atten-
tion, Action, Strategies, and Bottom-Up Constraints, Hauppauge,
NY: Nova Science.
Rowlands, M. (2005) ‘‘The Cognitive Penetrability of Perception,’’ in
A. Raftopoulos (ed.) Cognitive Penetrability of Perception: Atten-
tion, Action, Strategies, and Bottom-Up Constraints, Hauppauge,
NY: Nova Science.
Santangelo, V, Van der Lubbe, R. H., Belardinelli, M. O. and
Postma, A. (2006) ‘‘Spatial Attention Triggered by Unimodal,
Crossmodal, and Bimodal Exogenous Cues: A Comparison of
Reflexive Orienting Mechanisms,’’ Experimental Brain Research,
173(1): 40–8.
Santangelo, V., Olivetti Belardinelli, M., Spence, C. and Macaluso, E.
(2009) ‘‘Multisensory Interactions Between Voluntary and
COGNITIVE PENETRATION OF COLOUR EXPERIENCE 61

Stimulus-Driven Spatial Attention Mechanisms Across Sensory
Modalities,’’ Journal of Cognitive Neuroscience, 21: 2384–97.
Segal, S. J. (1971) ‘‘Processing of the Stimulus in Imagery and Percep-
tion’’, in S. J. Segal (ed.) Imagery: Current Cognitive Approaches,
New York: Academic Press.
—— (1972) ‘‘Assimilation of a Stimulus in the Construction of an
Image: The Perky Effect Revisited,’’ in P. W. Sheehan (ed.), The
Function and Nature of Imagery, New York and London: Academic
Press.
Segal, S. J. and Fusella, V. (1971) ‘‘Effects of Images in Six Sense Modal-
ities on Detection (d’) of Visual Signal from Noise,’’ Psychonomic
Science, 24: 55–56.
Sekuler, R., Sekuler, A. B. and Lau, R. (1997) ‘‘Sound Alters Visual
Motion Perception,’’ Nature, 385: 308.
Siegel, S. (2005) ‘‘Which Properties Are Represented in Perception?,’’
in T. Szabo Gendler and J. Hawthorne (eds.) Perceptual Experi-
ence, Oxford: Oxford University Press.
Spence, C. and Driver, J. (2004) (eds.) Crossmodal Space and Crossmo-
dal Attention, Oxford: Oxford University Press.
Sto
¨rmer, V. S., McDonald, J. J. and Hillyard, S. A. (2009) ‘‘Cross-
modal Cueing of Attention Alters Appearance and Early Cortical
Processing of Visual Stimuli,’’ Proceedings of the National Academy
of Sciences of the United States of America (PNAS), 106(52):
22456–61.
Thomas, N. J.T. (2008) ’’Mental Imagery’’, in E. N. Zalta (ed.) The Stan-
ford Encyclopedia of Philosophy (Winter 2008 Edition), URL =
<http://plato.stanford.edu/archives/win2008/entries/mental-imagery/>.
Winawer, J., Witthoft, N., Frank, M. C., Wu, L., Wade, A. R. and
Boroditsky, L. (2007) ‘‘Russian Blues Reveal the Effects of
Language on Color Discrimination,’’ Proceedings of the National
Academy of Sciences of the United States of America (PNAS),
104(19): 7780–5.
62 FIONA MACPHERSON