Susanna Siegel
(revised September 2009)
If you want to know whether there is any mustard in the refrigerator, it is a good idea to open the door and look. If you see the mustard, you can end up knowing its whereabouts: it’s in the fridge. If instead of looking for the mustard, you pictured the fridge interior in a daydream, and then relied on your daydream to confirm whether the fridge contained mustard or not, you wouldn’t end up knowing anything about the mustard or the fridge, but you might nevertheless end up with a true or false belief.
As the mustard example illustrates, it is common to regard perception as a special kind of input to belief that allows us to compare hypotheses with the world, so that we may assess whether those hypotheses are true. Even philosophers who were cautious about assigning perception more than a causal role in relation to knowledge regard perception as involving a special sort of input to the mind, different in kind from belief and judgment. For instance, Locke distinguished ideas of sensation from ideas of reflection, and Hume distinguished impressions from ideas. Both in common practice and in philosophy, perception is regarded as a distinctive kind of mental state that serves as an input to belief, and is distinct from it.
In this paper, I will argue that despite the differences between perception and belief, perception involves states that are importantly similar to beliefs: conscious visual perceptual experiences, where these include visual hallucinations that are introspectively indistinguishable from perceptions.[1] I argue that visual perceptual experiences share an important feature with beliefs: they have contents, in a sense to be explained. The discussion will be oriented around interpreting, developing and defending the following thesis, and then discussing its significance:
The Content View: All visual perceptual experiences have contents.
The kind of content at issue meets two constraints. Contents are true or false, and contents of experience are conveyed to the subject by her experience.[2] The sense in which experiences have contents (according to the Content View) thus picks up the strand of ordinary usage that takes contents to be things conveyed by sources of information (as when we speak of the contents of a newspaper story). Experience contents need not individuate experiences. We can, apparently, ostend experiences by using introspection or by describing the circumstances in which the experience is had, without first settling whether experiences have any contents at all, let alone which contents (if any) they have.
In
contemporary discussions the Content View is widely held, and is even
presupposed in many debates about perceptual experience.[3]
But some philosophers have objected to the Content View on the basis of Naive
Realism, claiming that Naïve Realism is the more commonsensical view, and in
one case finding precedent for it in
The discussion will proceed as follows. After developing the Content View and discussing the notion of accuracy in section 1, I will present the Argument from Accuracy in section 2. In section 3, I identify a flaw with this argument. In section 4, I present an argument for the Content View that corrects the flaw: the Argument from Appearing. In section 5, I consider two objections, each suggested by Charles Travis’s recent discussion of uses of ‘looks’ and its cognates, that are directed at the Argument from Accuracy, the Argument from Appearing, and the Content View itself. Finally, in section 6, I discuss the significance of the Content View, and its status vis à vis Naïve Realism.
The Content View can be refined into a proposal that finds the following similarity between visual perceptual experiences (hereafter, visual experiences or just plain experiences) and beliefs: like beliefs, maps and newspapers, visual experiences have contents, and just as the contents of beliefs describe conditions under which the belief-state is true, so the contents of experiences describe conditions under which the experience is accurate. According to this proposal, experiences are the kinds of states that can be accurate, and their contents describe conditions under which they have this status.
Just which accuracy conditions are contents of experience will be discussed in detail in the sections that follow. We can begin by focusing on accuracy itself.
Accuracy can be construed as acuity, or it can be construed as absence of error. When accuracy is construed as acuity, the greater the degree to which an experience is accurate, the greater acuity the perceiver has, so that having microscopic eyes of the sort Locke and Leibniz discussed would yield relatively more accurate experiences, whereas having normal human eyes would yield relatively less accurate experiences (Locke 1689, Book 2, Chapter 12, section 14).[5] In contrast, when accuracy is construed as freedom from error, merely omitting truths about a situation will not by itself result in an inaccurate experience.[6]
The two notions of accuracy yield different comparative verdicts about degrees of accuracy. The visual experience of someone legally blind might be so seriously degraded that when faced with someone wearing a purple shirt, all they see is a patch of purple. Compare this degraded experience to the experience of a sighted person who can see the person’s face, their gestures, and their surroundings. When accuracy is construed in terms of acuity, then the degraded experience is less accurate than the sighted person’s experience. When accuracy is construed as freedom from error, then both experiences may be equally accurate. The difference in comparative classifications of degrees of accuracy can also be illustrated by assuming for a moment that experiences have contents. Suppose there are three books on an otherwise empty table – two of them close together directly in view from the head of the table, the third at the far end but still (potentially) in view. Now consider a pair of experiences that differ in their contents:
Experience A: <there are two books on the table>
Experience B: <there are only two books on the table>
Experiences A and B, we can suppose, do not differ in acuity: the degree of resolution is the same both times. So if accuracy is acuity, then experiences A and B are accurate. But if accuracy is freedom from error, then experience B is less accurate than experience A.
The Content View construes accuracy as freedom from error. When accuracy is construed in this way, the idea that accuracy comes in degrees can be understood by considering separate respects (such as location, color, or shape) in which an experience is or is not accurate, and then breaking accuracy conditions down into separable contents. The Content View can thus respect the observation that accuracy comes in degrees while truth doesn’t. Completely accurate experiences would be ones in which all contents are true. Partly accurate experiences would be ones in which only some contents are true. For example, suppose experiences have contents, and consider an experience with the contents that there is a red cube in front of you, and a blue ball off to the left. If there really is a red cube in front of you and a ball off to the left, but the ball is black rather than blue, then your experience would be accurate with respect to the location of the ball, but inaccurate with respect to its color. It would be partly accurate, but inaccurate overall.
In the case of belief, there are two things that can be true or false: the belief state, and the belief contents. Contrast the case of hope. If you hope that winter ends soon, your hope itself is neither true nor false, but the content that winter ends soon is. (Here the notion of content is not tied to accuracy conditions of a state, but simply to something that can be true or false). According to the Content View, experiences are more like beliefs than hopes, in that they involve two things that can be free of error: the experience itself, and its content.
But in another way, experiences differ from beliefs. In the case of experiences, it is useful to be able to talk about degrees of their accuracy, whereas in the case of beliefs, it is not helpful to talk about degrees to which they are true. That is why it is better to define contents of experience as conditions on accuracy, rather than as conditions on truth, even though the notion of accuracy as freedom from error is similar to the notion of truth.
Given the construal of contents as accuracy conditions, we can distinguish between two aspects of experience that mirror the Fregean distinction between the force and sense of a sentence, and well as an analogous distinction for mental states between their function and content drawn by the 19th century German psychologists Carl Stümpf and Oswald Külpe in their discussions of perception and imagery (Dummett 1981).[7] To argue for the Content View, one has to show both that experiences (like hopes) involve relations to a relatum that can in turn be true or false, and also that experiences (unlike hopes) can themselves be accurate or inaccurate.
2. The Argument from
Accuracy
We can now examine the first argument aimed at supporting the Content View: the Argument from Accuracy.
The Argument from Accuracy
P1: All experiences are accurate or inaccurate.
P2: If all experiences are accurate or inaccurate, then all experiences have accuracy conditions.
Conclusion: All experiences have accuracy conditions.
Let us examine its premises in turn.
The case for P1
Accuracy and inaccuracy are properties a token experience would have relative to a situation: most naturally, the situation in which the experience is had. Later on, we will consider whether token experiences might be accurate or inaccurate with respect to other situations. For now we can take P1 to say that for all token experiences, there are some respects in which the experience is accurate or inaccurate relative to the situation in which it is had.
We can easily distinguish hallucinations and illusions from completely successful perceptions. Here are some examples:
Airport hallucination: You are at home, but hallucinating being in an airport.
Fishtank: The fish you are seeing is blue and it looks blue. It is at location L, but looks to be at location L*, which is a bit to the right of L.
Lunchtime: Behold your sandwich, cut in half on plate. It is as it looks.
Experiences such as Fishtank and Lunchtime commonly occur. Experiences like Airport hallucination probably never occur, yet we still find it easy to distinguish them from illusions and completely successful perceptions. Hallucination and illusion fall short of complete success: in hallucination, perceptual contact is missing; illusions are misleading guides to what is in the environment. In contrast, completely successful perceptions typically lead to knowledge.[8] Experiences in this last group are often called “veridical.”
A related distinction between kinds
of experiences is also natural for us to draw, even if we don’t have labels in
ordinary language that mark this difference. In that sense the distinction is
pre-theoretical. It is illustrated by contrasting Fishtank with another experience,
Fishtank2:
Fishtank: The fish you are seeing is blue and it looks blue. It is at location L, but looks to be at location L*, which is a bit to the right of L.
Fishtank 2: The fish you are seeing is blue and at location L, and it looks blue and looks to be at L.
In Fishtank, there need be no illusion with respect to the fish’s color or shape, but there is an illusion with respect to location. The experience is veridical with respect to color and shape, but not veridical with respect to location. It is thus partly but not completely veridical. In contrast, Fishtank 2 is completely veridical, so far as this experience is described.
Interestingly, hallucinations too can be veridical in some respects, and can differ from one another their degree of veridicality:
Airport: You are at home, but hallucinating being in an airport.
Amazing Coincidence: Your experience is just as it is now, from your point of view, but you are hallucinating. But the scene before your eyes is exactly as presented in your hallucination.
In one sense of ‘veridical’, Airport is less veridical than Amazing Coincidence. But this sense differs from the sense of ‘veridical’ used to label completely successful perceptions, as distinct from hallucinations and illusions, since Airport and Amazing Coincidence are both hallucinations.
We can label the two uses of “veridical” as follows. Sometimes it is used to denote experiences that are veridical of the things seen. Call these experiences ‘strongly veridical.’ Hallucinations cannot be strongly veridical. Other times, “veridical” is used to describe experiences that are veridical without being veridical of any object that is seen. For instance, through an amazing coincidence, a hallucination could occur in the presence of exactly the sort of scene that is hallucinated. Call “weakly veridical experience” experiences that are veridical, whether or not they are strongly veridical.
Strongly
veridical experiences may fall short of being completely successful. Completely successful experiences are best thought of
as experiences in which the subject perceives both an object and its
properties. Consider Simone, who systematically misperceives green things, so
that to her green things look yellow. Suppose that by stimulating Simone’s
brain area V1 while she looks at a green cube, you accidentally induced in her
an experience as of seeing a green cube, when otherwise she would have an
experience as of seeing a yellow cube. The intervention does not correct
Simone’s systematic error. But on the basis of the experience that the
intervention helps produce, it would be natural for Simone to form the true
belief that there is a green cube before her. Intuitively, Simone’s experience
is not completely successful, because the greenness of the cube does not play
the right role in producing the experience. Simone perceives the cube, and it
looks green to her, but she does not perceive its greenness.
Cases like this one are called
‘veridical illusion’.[9] They suggest that leading to true belief does
not seem to be enough for complete success. In general, an experience will be
completely successful if it is a case of seeing o when o looks F, o is F, and o’s looking F is due to o’s
F-ness, and not to any irregular intervention. These experiences are not just
strongly veridical, they are superstrongly veridical.
They are cases of optimal perceptual contact with an object and select
properties.
With the distinction between strong and weak veridicality in hand, we can see that the pre-theoretical distinction that was illustrated earlier between degrees of veridicality is not sensitive to whether experiences are strongly veridical. It is a distinction between degrees of weak veridicality. According to the straightforward account of what underlies our classification of experiences into partial and complete veridicality, weak veridicality is accuracy, and degrees of weak veridicality are degrees of accuracy. In the Amazing Coincidence, the experience is accurate with the respect to the situation that the hallucinator is actually in, even though the hallucinator is not perceiving anything in that situation. If the experience in the Amazing Coincidence represented that you were in a room containing yellow chairs, when in fact you were hallucinating while standing in a room containing red chairs, it would not be completely weakly veridical, but it could still be weakly veridical to a high degree.
The previous discussion suggests the following defense of P1: When given certain descriptions pairing token experiences with situations in which they are had, such as Fishtank, Fishtank2, Lunchtime, Airport, and Amazing Coincidence, we easily classify them into completely falsidical, partly veridical, or completely veridical (at least, completely veridical, for all the descriptions specify), and the best explanation of these classifications is that the experiences classified as completely veridical are completely accurate (so far as the descriptions specify), and experiences classified as partly veridical are only partly accurate.
What else might explain our classifications?
The abductive defense of P1 just described will not succeed if there are alternative, superior explanations of what underlies comparative classifications of Fishtank and Fishtank 2, Airport and Amazing Coincidence, and our classification of some of these (Fishtank) as partly veridical, others (Lunchtime, Fishtank 2 and Amazing Coincidence) as completely veridical, and still others (Airport) as completely falsidical. The straightforward account has simplicity on its side. But let us consider the alternatives.
One might try to argue that the only classifications of accuracy in the vicinity are classifications of judgments downstream of experiences. This alternative will bring us straight to the heart of the controversy over the Content View, and it will be useful to have the second argument for the Content View on the table before exploring it. The second argument – the Argument from Appearing – will be presented in section 4, and we’ll return to the alternative explanation of our comparative classifications of experiences in section 5.
According to another alternative, the classifications are sensitive to a notion of weak veridicality that does not involve the notion of accuracy, but instead involves the notion of indiscriminability from strongly veridical experiences. Such a notion of weak veridicality might be defined as follows:
An experience is weakly veridical iff it is indiscriminable from a strongly
veridical experience of perceiving something that has a cluster of properties F, and the experience is had by someone in the presence of something that has that cluster of properties.[10]
If properties are in a cluster, then if they are instantiated, they are all instantiated by the same thing. The cluster of properties F would have to be such that any two experiences that presented something as having F would be indistinguishable to the subject. An experience that presents something as being F is strongly veridical, in turn, just in case it provides a basis for knowing that something is F. Let us assume that ‘providing a basis for knowing’ can be spelled out in some acceptable way. Veridicality would then be partial if something is present with only some of the properties in cluster; it would be complete if something is present with all the properties in the cluster.
Ultimately, this proposal does not seem to succeed in avoiding the notion of accuracy. Its right-hand-side relies on the idea that there is a cluster or properties that an experience can either match or mismatch. According to the proposal, the status of an experience as weakly veridical depends on whether the cluster of properties that figures in the indiscriminability property is instantiated. An experience will be weakly veridical if and only if a cluster of properties is instantiated that matches the properties in the indiscriminability. Little daylight can be found between this kind of matching and accuracy. Through the dependence of weak veridicality on this kind of match, the proposal seems to reintroduce the notion of accuracy, rather than providing an alternative to it.[11]
The case for P2
So far, we
have defended the first of two premises in the Argument from Accuracy. The
second premise is P2:
P2: If all experiences are accurate or inaccurate, then all experiences have
accuracy conditions.
P2 should be uncontroversial. If something is accurate, then there is something else in relation to which it is accurate. If a map is accurate, then there is some spatial area in relation to which it is accurate. If a story about Simone is accurate in some respects but not others, then Simone has some of the features attributed to it by the story but not others. When a map, a story, a mental state, or anything else is accurate, there is some situation of which it is accurate. Attributing accuracy to something thus involves comparing it to something else.
Let us focus on the case of token experiences (though the following points hold for maps, stories and anything else that can be accurate or inaccurate as well). If an experience is accurate, then there are some conditions under which it is accurate. The conditions obtain in the situation with which we are comparing the situation when we say that it is accurate. If an experience is inaccurate with respect to that situation, then there is some mismatch between the experience and the situation. If there were no mismatch, the experience would be accurate. The conditions in which there is no mismatch are accuracy conditions. Analogous points hold for maps, stories, and other mental states besides experiences that can be accurate or inaccurate.
When we categorize token experiences as veridical (or veridical in certain respects), we are comparing them to the situation in which they are had. If these classifications involve accuracy, then when we pre-theoretically categorize an experience as veridical, we are saying that it is accurate with respect to the situation in which it is had.
3. A flaw in the
Argument from Accuracy
One might think that P2 is trivial on the grounds that some claims analogous to it are clearly trivial. Consider an analog of P2 for blueness: If all objects are blue or not-blue, then all objects have blueness-conditions. A blueness-condition is a condition that obtains exactly when something is blue. Turning from blueness and blueness-conditions to accuracy and accuracy-conditions, if I see a red cube and my experience of seeing it is accurate, then there is trivially a condition that obtains exactly when my experience is accurate: namely, the condition that my experience is accurate.
If P2 is trivial, then all the weight in the Argument from Accuracy rests with P1, which will in turn be uncomfortably close to the conclusion - as it is just a trivial consequence away. By itself, the triviality of P2 would not be a flaw, as it would be compatible with the status of P2 as trivial that it is part of a nice simple argument that puts argumentative weight exactly where it should be – with the claim P1, that all visual perceptual experiences are accurate or inaccurate. Such an argument would not have much internal dialectical force, but it would bring into focus that the main action in defending an important consequence of the Content View – the claim that experiences have accuracy conditions –– lies in the defense of P1.
Considered as a defense of the Content View, however, there is a flaw with the Argument from Accuracy that arises from the status of P2 as trivial. The flaw is that the argument does not tell us what makes the accuracy conditions had by experiences suitable for being contents of experience, given the constraint that contents are conveyed to the subject by her experience. What needs support is not just the generic conclusion that experiences have any old accuracy conditions, but the more robust conclusion that they have accuracy conditions that are clearly fit to be contents of experience. The triviality of P2 shows that a transition from the generic conclusion to the Content View is not warranted. So the Argument from Accuracy is not an argument for the Content View as contents have been defined here. A good argument for the Content View should tie together accuracy conditions and contents, elucidating why experiences are assessable for accuracy in the first place, and how they could convey their accuracy conditions to the subject.
4. The Argument from
Appearing
We can find such elucidation in the Argument from Appearing. It attempts to unearth the idea that experiences are assessable for accuracy, aiming to explain what gives them accuracy conditions that are suitable for being contents.
The Argument from Appearing proceeds from premises about the phenomenal character of visual perceptual experience. The accuracy conditions that figure in its conclusion derive from the properties that are presented in visual phenomenology. Premise (i) claims that properties are presented in visual phenomenology, premise (ii) links these properties to instantiation at a world, which is in turn linked to accuracy conditions in premise (iii).
The Argument from Appearing
Premise (i)
All visual perceptual experiences present clusters
of properties as being instantiated.
Premise (ii)
If an experience E
presents a cluster of properties F as being instantiated, then:
Necessarily: things
are the way E presents them only if property-cluster F is instantiated.
Premise (iii)
If necessarily: things are the way E
presents them only if property-cluster F is instantiated, then:
there is a set of accuracy
conditions for E such that:
the conditions are satisfied in
a world only if there is something that has F in that world.
Conclusion 1: All visual perceptual experiences have
accuracy conditions.
Conclusion 2: All visual perceptual experiences have
contents.
Premise (i)
According to Premise (i), all visual perceptual experiences present clusters of
properties as being instantiated. Is this description of experiences phenomenologically apt?
Typically, our visual perceptual experiences are
cases of seeing objects, where the category of objects includes the sky as well
as ordinary objects such as cars, cups and pencils.[12]
Why think that properties are presented in such experiences? Consider the claim (made sometimes in
discussions of metaphysics) that there is no such thing as a ‘bare particular’:
that is, an object shorn of all of its properties.[13]
Premise (i) is motivated by
the idea that it is not possible for us to represent objects as so shorn in our
visual experience. When we see (or even when we merely seem to see) ordinary
objects, such as a cube, bare particulars do not figure in visual phenomenology
in any way. Properties enter the picture as well. For you to see a cube at all,
it must be part of your visual phenomenology that the cube has certain
properties: as it might be, having a certain number of facing edges and
surfaces, having a certain color, location, and so on.
Most of the time, visual phenomenology takes a stand
on which objects instantiate clusters of properties, both at a time and over
time.[14]
For instance, when you see a bird flying by, it looks as if a single object is
moving. Your experience does not remain neutral on whether it is the same
object at various points along the trajectory. But in some cases, an experience
may present the perceiver with a property without specifying what is
instantiating it. For instance, Dretske 1999
discusses the case in which you can’t tell whether the moving train is the one
you’re seeing or the one you’re sitting in. Even here, though, this is part of
a visual experience that does attribute properties to objects.[15]
We thus see objects, and we can’t seem to see them
without our experience presenting them as having certain properties. But
more can be said about how properties figure in experience.
Suppose you see a
cube, and it looks red and cubical. Here your experience presents it as being
the case that there is a red cube before you. Contrast a hope that there is a
red cube in front of you. The properties of being red and cubical figure in the
content of the hope. But in hoping that there is a red cube in front of you, it
need not be presented to you as being the case that there is a red cube in
front of you. To make this vivid, suppose your eyes are closed, you’re not
holding onto anything, and not engaging in any visual imagination. Under such
circumstances you could still hope that when you open your eyes there will be a
red cube in front of you. In contrast, when such a property cluster (redness, cubicality, and being nearby) figures in visual perceptual
experience, the experience presents it as being the case that a red cube is
nearby. It is the fact that properties figure in this way that will eventually allow
us to draw the link to accuracy conditions.
These considerations about the kind of visual phenomenology involved in seeing ordinary objects support premise (i), and they apply equally to cases of merely seeming to see objects.[16] The same considerations also suggest that a sense of ‘looks’ and ‘appears’ can be defined in which when you see an object, it looks (appears) to you to have properties. (We will revisit this suggestion for defining a sense of ‘looks’ in section 5).
It
should be taken as analytic that if an experience presents a property as being
instantiated, then it presents the property as being instantiated by something
other than the experience itself. There may be more to the phenomenal character
of an experience than what it presents. For instance, premise (i) allows that experiences present themselves as having non-relational
properties such as being blurred. On some accounts of blurred vision (e.g.,
Smith 2008, Pace 2007), part of what is presented (in a broader sense than the
one used here) to the subject by her blurred experience is that her vision is
blurred. Our more restrictive notion is better suited to be linked eventually
with accuracy conditions, and is compatible with treatments of blurry vision
such as Smith’s.
Premise
(ii)
According to premise (ii),
If an experience E
presents a cluster of properties F as being instantiated, then:
Necessarily: things
are the way E presents them only if property-cluster F is instantiated.
Premise
(ii) is an instance of a more general claim about presentation: if a state
presents such-and-such as being the case, then things are the way the state
presents them only if such-and-such.
The general claim seems plausible,
no matter what presentation is. But to bring the general claim into focus, it
may be useful to discuss the notion of presenting such-and-such as being the case
a bit further.[17] Other mental states
besides experiences can present things as being the case, and some states can
do this, without involving any phenomenal character (though (ii) would be just
as plausible, even if presentation were tied specifically to phenomenal
character). For instance, belief and supposition are modes in which things may
be presented as being the case. The kind of commitment involved in belief is a
specific kind of presentation, but not the only kind, as it is missing in
supposition and imagination. For instance, my supposition that it will not rain
tomorrow presents it as being the case that it will not rain tomorrow, but my
supposition does not involve any kind of commitment.[18]
Premise (iii)
If necessarily: things are the way E presents
them only if property-cluster F is instantiated, then:
there is a set of accuracy
conditions for E such that:
the conditions are satisfied in
a world only if there is something that has F in that world.
Like
premise (ii), premise (iii) is an instance of a general claim that is
independent of any claims about phenomenology:
If things are the way that a state X present them as
being only if such-and-such,
then X has accuracy conditions
that are satisfied in a world only if such-and-such.
For instance, when visual phenomenology presents an object o as having F (where o is seen), something’s having F (maybe o) is a condition on the accuracy of experience.
Someone might worry that this defense of premise (iii) lets in the sort of trivial accuracy-condition that P2 of the Argument from Accuracy allowed. But premise (iii) excludes such accuracy conditions. We can see this by examining the kind of accuracy conditions it describes, using variants of cases described earlier.
Fully Falsidical Hallucination: You are at home, but are hallucinating being in an
airport.
Fully Amazing Coincidence: You hallucinate being in an airport, and you are in an airport. The scene before your eyes is exactly as presented in your hallucination.
Less
Amazing (but still pretty amazing) Coincidence: You hallucinate being in an
airport, and you are in an airport that is exactly like the one you’re
hallucinating in all some but not all respects.
These cases describe pairs of hallucinations and situations
in which they are had. But we can also fix on the first hallucination (Fully Falsidical Hallucination)
as an anchor point, and think of the other cases as counterfactual
situations relative to which we can evaluate the original hallucination.
Relative to the Less Amazing situation, the hallucination would be more
veridical than it is in the original case, since all three experiences are
hallucinations of being in an airport. Relative to the Fully Amazing situation,
it would be still more veridical.
We reach these verdicts concerning
veridicality on the basis of information about what conditions obtain in the
counterfactual situation. In general, suppose we evaluate an experience (such
as Fully Falsidical
Hallucination) with respect to a situation other than the one in which it
is had. Call this situation world w. We will count the experience as veridical
with respect to w only when properties are instantiated in w that are presented
in the experience.[19]
These conditions are thus veridicality conditions that can be satisfied in
worlds where the experience does not occur. And if veridicality conditions are
accuracy conditions, then these accuracy conditions don’t include the trivial
condition that the experience is accurate, as they do not require the presence
of the experience at all. So these
accuracy-conditions are not analogous to the trivial blueness-conditions
discussed in section 3.
From
Conclusion 1 to Conclusion 2
In the Argument from Appearing,
Conclusion 2 as well as Conclusion 1 is needed because given the notion of
content in the Content View, contents, unlike mere accuracy conditions, are
conveyed to the subject by her experience.
We can distinguish between three ways in which a content
can be conveyed to the subject by her experience. First, a
content is conveyed by experience, if it would be a content of explicit
beliefs that are natural to form on the basis of visual experience.[20]
Second, a content is conveyed to the subject by her
experience if it enables the experience to guide bodily actions. For instance,
suppose you see the door but don’t form any explicit beliefs about the shape of
its doorknob, yet you adjust your grip in advance of touching the doorknob as
you reach for it. This could be a case of visual experience guiding action.
Finally, a content C is conveyed to the subject by her experience if it is
manifest to introspection that C is a content of experience.
Unlike the Argument from Accuracy, the Argument from Appearing defends the idea that experiences have accuracy conditions that are fit to be contents. It does this in two ways. First, as we’ve seen, the kind of accuracy conditions described in premise (iii) exclude the trivial accuracy conditions of the sort allowed in by P2 of the Argument from Accuracy, which are not conveyed to the subject in any of the three ways listed above. Second, if there are properties presented in visual phenomenology, this opens the possibility that since those properties are conveyed to the subject, the accuracy conditions they directly determine also so conveyed.
What would it be for a property presented in experience to be conveyed to the subject by her experience? In cases of seeing objects, properties that are presented in visual phenomenology are properties that objects look to the perceiver to have when she sees them. Such properties can be conveyed to the subject, in the same ways that the contents of an experience can be conveyed to a subject. And there is good reason to think that such properties are conveyed in these ways. For instance, upon seeing the banana, it is natural to believe that it is yellow and bent, and this is arguably because those properties (being yellow, being bent) are presented in visual phenomenology.[21] Similarly, upon seeing the doorknob, it is natural for one’s active movements to adjust to what one sees, as when you automatically adjust your grip to match its shape.[22] Finally, it seems manifest to introspection that visual phenomenology presents spatial properties (such as being nearby or in front of the perceiver), color properties (or properties closely related to colors), shape and luminance properties - though it is doubtful that for every property presented in visual phenomenology, it is manifest to introspection that it is so represented.[23]
Supposing that properties presented
in experience are conveyed to the subject, does this support the idea that
accuracy conditions that derive from those properties are so conveyed? The
accuracy conditions described in premise (iii) derive directly from properties,
so it is hard to see how accuracy could fail to be conveyed to the subject in
whatever way properties are. If it is natural to believe that the banana is
yellow when it looks yellow because the property of yellowness is conveyed to
the subject by her experience, then the content that something is yellow will
be conveyed as well. Likewise, if the property F-ness is presented in
experience guides one’s action, then this seems enough for the content that
something is F to guide one’s action.
Only if vs. iff accuracy conditions
The ultimate characterization of accuracy conditions
in premise (iii)’s consequent is an only-if
condition. It is plausible that
whenever there are some conditions C such that X is accurate only if C, there
are some (perhaps stronger) conditions C* such that X is accurate iff C*. But the
argument from appearing does not tell us what the full accuracy-conditions C*
are, and one might worry that even if C meets the further constraints on being
a content of experience, C* will not. So it would be more illuminating if “only
if” in (ii) and (iii) could be replaced by “iff”,
resulting in a full definition of accuracy conditions in terms of properties.
Premise (iii) remains plausible once ‘only if’ is
replaced by ‘iff’. But to make this replacement,
‘only if’ would have to be replaced by ‘iff’ in
premise (ii) as well. A version of premise (ii) with ‘iff’
would be true only given the assumption that what an experience E presents as
being the case is exhausted by E’s presenting certain properties as
instantiated. And here complications arise. One such complication concerns the
role of objects that are seen, and specifically whether such objects are part
of an experience’s accuracy conditions. Suppose you see Franco, and your
experience represents him as sitting down. In order for the experience to be
accurate with respect to a world, does Franco himself have to be sitting down
in that world, or is it enough for accuracy if a qualitative duplicate of Franco
is F is that situation? For instance, is your experience veridical with respect
to a world where Franco’s twin is sitting down but Franco is standing up? As
they stand, the ‘only if’ versions of (ii) and (iii) leave unsettled whether
accuracy conditions track objects seen across worlds, since they do not specify
which object has to instantiate the cluster of properties F, in order for the
experience of seeing that object to be accurate. To get a full definition of
accuracy conditions, this issue must be settled.
But whichever way the issue is settled, the resulting accuracy conditions are fit to be contents of experience. If you see Franco and your experience represents him as sitting down, it is natural to believe on the basis of your experience that Franco is sitting down. It is also natural to believe that someone with a certain appearance is sitting down. These are both ways for contents to be conveyed to a subject. So both options result in accuracy conditions that have a good claim to being conveyed to the subject by her experience.
Why properties? Fregean contents and centered
worlds
The assumption that what an experience E presents as being the case is exhausted by E’s presenting certain properties as instantiated is called into question by scenarios in which it seems prima facie that two perceivers accurately represent different properties, yet have experiences that are phenomenally the same. Objects typically look to stand in certain spatial relations to the perceiver, such as being nearby or within reach. On some views (e.g. Egan 2006 and this volume), this involves the presentation of "centering features" defined in terms of evaluation in centered worlds, where these are not properties. For example, in contrast to being nearby Susanna, which is a property, the centering feature being nearby the center is not a property, since being a center is merely a formal feature of a centered world. The claim that centering features are presented in experience is motivated by the idea that pairs of veridical phenomenally identical experiences can nonetheless be associated with different locations.
Centering features provide a level at which such phenomenally identical experiences present the same thing. Fregean modes of presentation have been invoked to play a similar role in response to inversion scenarios. In spectral inversion, phenomenally identical pairs of veridical color experiences are associated with different color properties. For instance, in one such scenario, Invert and Nonvert’s color experiences are phenomenally the same, but Invert’s experience presents red while Nonvert’s experience presents green. Is the phenomenal similarity between Invert and Nonvert a mere ‘raw feel’, or does it have some other status? Chalmers (2004) argues that Invert and Nonvert’s shared phenomenal character co-varies with a level of content that is composed of Fregean modes of presentation, where these are part of a two-dimensionalist theory of experience content. (Thompson (forthcoming) defends a similar two-dimensionalist theory on the basis of inversion scenarios involving spatial properties).
Both the two-dimensionalist Fregean theory of experience content and the theory that experiences present centering features suggest an objection to premise (i): when you see objects, they look to have features that are not properties, such as centering features or modes of presentation.[24] However, this objection to premise (i) can be met, since centering features and modes of presentation could be presented in experience along with the location and color properties with which they are associated. Indeed, each of these theoretical devices is invoked to explain the means by which such properties are presented in experience. Premise (i) does not entail that any pair of phenomenally identical experiences present exactly the same properties. In addition, both the view that experiences present centering features and the Fregean views about experience are versions of the Content View, and so do not ultimately challenge the conclusion of the Argument from Appearing.[25]
Centering features and Fregean contents also pose a challenge to the move from only-if to iff accuracy conditions. As we noted earlier, a version of premise (ii) with ‘iff’ would be true only given the assumption that what an experience E presents as being the case is exhausted by E’s presenting certain properties as instantiated. But this assumption is false if experience presents centering features rather than properties, or in addition to them. A modified version of the argument could deal with this issue, however, by replacing references to properties by references to features (properties or centering features), and by replacing references to worlds by reference to centered worlds. So this obstacle to defining ‘iff’ accuracy conditions is easily overcome.
A similar dialectic surrounds two-dimensionalist Fregean views. On
these views, even if the contents of an experience E derive from what E
presents as being the case, what experiences present as being the case is not
exhausted by the instantiation of properties, because properties are presented
under a mode of presentation that can pick out different properties in
different worlds. For instance, according to Chalmers and Thompson, the mode of
presentation for redness (roughly ‘the property normally causing reddish
experiences in me’) picks out different properties in different worlds. These
theories allow that two experiences could present redness as being
instantiated, yet differ in the mode of presentation of redness, and hence in
their accuracy conditions.
It is possible to formulate biconditional versions of (ii) and (iii) that would accommodate these two-dimensionalist Fregean views, on which accuracy conditions are determined by modes of presentation of properties.[26] In effect these theories posit two sets of accuracy conditions for experience: one set that co-varies with phenomenal character (found at the level of sense or modes of presentation), the other set that does not (found at the level of reference or properties presented in experience). When coupled with two-dimensional theories of belief, both sets of accuracy conditions in experience will be conveyed to the subject, to the extent that they each have an analog in the contents of beliefs formed on the basis of experience.[27]
5. Two objections
from “looks”, “appears” and their cognates
Premise (i) of the Argument from Appearing, like the descriptions given earlier of Airport, Fishtank and Fishtank 2, Lunchtime, etc., all use “look”, “present”, or cognates. Without relying on descriptions like these, the argument could not get off the ground. I will now consider a pair of objections to such uses of ‘look’ and its cognates. The second objection in the pair brings us to the heart of the controversy over the Content View.
According to the first objection, there are no natural uses of English words “look”, “visually present” or their cognates that pick out contents of experience exclusively. Call this the semantic objection:
Semantic objection: No actual uses of ‘looks’ (or ‘looks F’) and its cognates in ordinary English exclusively track what is presented in experience.
The discussion so far has relied on the idea that we can use ordinary English expressions (including such locutions as looks F) to identify visual perceptual experiences, as opposed to mental states further downstream of perception. According to a specific version of the semantic objection, the only mental states that can be picked out by ordinary English uses of “looks F” and “looks to be F” are judgments that would be reasonable to make on the basis of experience. If that is true, then the putative descriptions of experiences used in arguing for P1 of the Argument from Accuracy and in discussing premise (iii) of the Argument from Appearing are defective. Similarly, the Argument from Appearing uses the phrase “experience presents clusters of properties”, and in cases where the experience is a case of seeing an object, the properties presented are meant to be properties that the object looks to the perceiver to have. So the Argument from Appearing depends on the idea that objects look to have properties – and here again the English expressions “looks F” and “looks to be F” are indispensable.
The semantic objection leaves open that there could be a special, regimented use of ‘looks’ that does exclusively track the contents of experience, yet bears enough resemblance to ordinary uses to make it reasonable to choose the English word for that purpose. [28] For all the semantic objection says, it could be such a regimented use that gives sense to the Argument from Appearing when it uses the notion of visual phenomenology presenting properties to a subject. We can thus distinguish the semantic objection from a more powerful objection, according to which no such regimentation is possible. Call this the psychological objection:
Psychological
objection: There is no mental state that uses of ‘looks F’ and its cognates
could track, other than judgments that would be reasonable to make on the basis
of experience.
If the psychological objection stands, then we can explain why we make the comparative classifications of veridicality in cases such as Fishtank, Airport, and the other cases described in section 2 without relying on the idea that experiences themselves are assessable for accuracy. Rebutting this objection thus contributes to defending the claim that the best explanation for our classifications is that experiences themselves can be accurate or inaccurate.
Something close to the semantic objection seems to be in play in Travis 2004. Travis raises doubts that any actual uses of ‘looks’ in English report contents of visual perceptual experience. His official target is the idea that “the representational content of an experience can be read off of the way, in it, things looked.” He says he will “begin to examine that idea by distinguishing and exploring two different notions of looks,” and that “[n]either…makes room for it.” (69). [29]
The first notion of looks is
characterized by Travis as follows:
On the first notion, something looks thus-and-so, or like such-and-such, where it looks the way such-and-such….does (would, might look. On this notion, Pia may look…like…her sister….That man on the bench looks old. (He looks the way an old man would, or might).[30]
In a footnote, Travis makes explicit that public looks – which can be expressed by the locutions “looking like X” or “looking F” - are supposed to contrast with looking like X or looking F to a perceiver.[31] Travis sometimes calls looks on the first construal ‘demonstrable looks’.[32] His main criticism is that demonstrable looks don’t fix on any single way the world has to be, in order for an experience to be accurate. Travis takes the question facing his opponent to be: what way does something have to be, in order to be the way that it demonstrably looks?[33] If there are conflicting ways a thing could be in order to be the way it demonstrably looked, then, Travis concludes, contents of experience cannot be read off of demonstrable looks.
Travis thinks this is just what we find. [34] A lemon, a lemon-shaped and colored soap, a small football in a lemon disguise, and countless other things could all share a demonstrable lemony look. How does the lemon (or the disguised football, etc) have to be in order to be the way it looks? You might think the answer is: it has to be yellowish and roughly lemon-shaped. But there are many lemony demonstrable looks, not all of which involve being yellowish and lemon-shaped. Maybe the lemon is cut in half. Maybe it’s got a green patch. So a lemony demonstrable look seems too coarsely grained an item to determine any set of contents for experience. Alternatively, you might think the answer is: it has to be a lemon. But that may seem arbitrary: why a lemon, as opposed to a football in disguise, or a well-crafted yellow soap? Even if we fix on a specific lemony demonstrable look that pins down shape, color and illumination, we still don’t seem to fix on any set of truth-conditions. Finally, you might think the answer is: once you fix a specific lemony demonstrable look, for something to be the way it demonstrably looks, it has to have just those properties that are involved in fixing the specific lemony demonstrable look: as it might be, lemon-shaped, roughly textured, yellowish, and so on.
Of these three answers to Travis’s question, the third seems the most sensible, as far it goes. But whatever status the answers may have, the question itself seems flawed, driven as it is by the idea that demonstrable looks might fix contents of experience. If the fact that a lemon demonstrably looks lemony doesn’t entail that it looks lemony to S, why should we think that the lemon’s demonstrable looks fixes the facts about the contents of S’s experience when she sees the lemon? At best these facts are fixed by S’s experience somehow picking up the demonstrable look of the lemon, when she sees it. But with the notion of picking up on a demonstrable look, we’ve introduced another kind of looking altogether. Demonstrable looks are irrelevant to fixing the content of experience. Since they are public, they are part of the way the world is, and as such don’t automatically determine how the world looks, appears, seems, or is presented to a perceiver. If any notion of looking is going to constrain the contents of experience, it must be looking some way to a perceiver.[35]
Suppose we grant that ‘looks’ and its cognates as actually used in English do not exclusively track what’s presented in visual phenomenology. This could be true, even if there was a special, regimented sense of ‘looks’ that did track what’s presented in visual phenomenology. To get from the semantic objection to the psychological objection, what’s needed is a reason to think that visual perceptual experience never presented objects as having properties. If it did, then a special sense of ‘look’ or its cognates (perhaps a regimented rather than an ordinary use) could just be assigned to label this aspect.
What structure would the phenomenal character of a visual perceptual experience have to have, in order for the objection to hold? There seem to be two answers:
Answer 1: Visual phenomenology is a pure raw feel, or Reidian sensation.
Answer 2: When a visual experience is non-hallucinatory, its visual phenomenology consists in the subject’s perceiving something other than properties.
According to Answer 2, in non-hallucinatory experiences, we perceive entities that are concrete and worldly. If those entities are objects we perceive and some of their properties, then this will allow us to define a sense of ‘looks F’ that picks out the properties objects look to us to have. This sense could then figure in the Argument from Appearing, and in the descriptions of Fishtank, Lunchtime, and the other experiences we have discussed. So this model of visual experience does not support the psychological objection, which denies that any special use of ‘looks’ could be defined that would exclusively track what’s presented in visual phenomenology. To support the psychological objection, it has to be the case that properties are never presented in visual phenomenology. Travis seems to endorse this conclusion. He writes:
"Perception can…make the world bear on what one is to think by furnishing access to things being as they are. Insofar as things being as they are is a different candidate object of perception than A being F,G, H,… then that's a reason not to think that perception or its phenomenal character (whatever that is) involves a commitment to the truth of some proposition." (p. 20, unpublished, quoted with permission.)
According to Travis, non-hallucinatory experiences are perceptual relations to ‘things being as they are’, and things being as they are differs from objects (or anything) having properties, and from anything individuated by objects and the properties they instantiate. The phenomenal character of non-hallucinatory experiences consists in this relation.
Travis’s position is a version of what we can call Radical Naïve Realism. According to Radical Naïve Realism, all non-hallucinatory experiences consist in a perceptual relation to something other than properties.
Radical Naïve
Realism: All non-hallucinatory experiences consist in a perceptual relation to
a worldly item, and properties are not among the things the subject is
perceptually related to.
Both Answer 1 (the raw-feel view) and Answer 2 (Radical Naïve Realism) deny that properties ever figure in experiences. (Answer 1 says the same about objects.) Phenomenologically this is highly dubious. Normally, when we see objects, we can discern where they are in relation to us, which bits of space they occupy, and in this informational feat visual phenomenology does not seem to be merely incidental. Visual phenomenology changes with big changes in what perception furnishes us access to. If we see a teapot in one case and a writhing snake in another, then the specific conscious character of each experience differs, as do the features of the world that to which the experience gives us access. Within a Naïve Realist framework, we need properties to specify which aspects of the experience we pick up on in experience, and correlatively which specific phenomenal character an experience has. Radical Naïve Realism denies, implausibly, that experience presents us with properties of the things we see.
Radical Naïve Realism is radical in two ways. First, as we just noted, its construal of the phenomenal character of experience leaves out reference to properties but not to objects. This construal might be suitable if objects ever looked to us like bare particulars, but they don’t. Second, Radical Naïve Realism denies the intuition that something is missing from cases of predicative veridical illusion, such as the case of Simone (2.1). According to Radical Naïve Realism, we never perceive properties (or property-instances) of the things we see, so there is nothing short of complete successful perceptual contact when blue things look blue to Simone, but only due to an intervention that removes a color illusion to which she would otherwise be subject if left to her own devices. Radical Naïve Realism denies that there are any superstrongly veridical experiences. In contrast, standard Naïve Realism takes superstrongly veridical experiences to be the central case of visual perceptual experience from which theorizing should proceed. We’ll soon see some examples.
6. The Significance of the Content View
We have seen that the Content View can be resisted by denying that properties are presented in experience. We can also ask: is the Content View unavoidable, so long as properties are presented in experience? If so, this would be a powerful philosophical result. There are many ways in which properties could be presented in experience. In particular, properties are presented in experience, even according to theories that are traditionally taken to be at odds with the idea that experiences have contents, such as Naïve Realism as it is standardly construed. If the Content View is unavoidable given widely accepted assumptions, then it will not be a parochial thesis, of interest only in a small corner of philosophy, but will rather be a thesis with wide application that can be used in the analysis of perception across a wide range of theoretical assumptions. Just as important questions about belief are usefully posed within a framework assuming that beliefs have accuracy conditions, the same may be true of perception.
One way for properties to be presented in experience leads directly to the Content View. Call this way the Property View.
Property View: All experiences involve relations to properties presented in experience, and are accurate only if those properties are instantiated.[36]
Since the Property View says that experiences have accuracy conditions that derive from the properties presented in experience, it leads to the Content View (given the assumption defended earlier that these properties are conveyed to the subject). Since the Property View entails the Content View, to oppose the Content View, one must also oppose the Property View.
The Property View may be resisted in several ways. First, properties might figure in experience in a way that avoids the Property View. For instance, according to classical sense-datum theories of the sort defended by Russell (1912), experiences consist in perceptions of sense-data (construed as mental objects) and their properties. The properties had by sense-data were thought to be different from properties of external objects, but systematically related to them. For instance, whereas apples are red, the sense-data you have when seeing an apple is red’. And since sense-data were thought to be located in mental space (rather than in the same space as the external apple), they couldn’t have exactly the same spatial properties (notably depth) as the external objects that were thought to cause them.[37] Second, Reidian sensations, or ‘raw feels’ involve properties to the extent that they purport to provide qualitative types of experience (so that two perceivers, or the same perceiver in different situations could have experiences with the same raw feel). Since these properties are not presented in experience as properties instantiated by anything at all, the Property View is avoided.
The pure raw feel view and the classical sense-datum theory each face substantial objections. First, according to the raw feel view, neither objects nor properties are presented in experience, which leaves it mysterious what role experience plays in enabling the subject to distinguish objects from one another, and figure from ground. Second, the raw feel view allows that two experiences could be phenomenally the same, while varying enormously in which contents would be natural to believe on the basis of the experience. Finally, the classical sense-datum theory faces the challenge of making sense of the notion of mental space to house sense-data, where mental space is distinct from the space in which our bodies and other external bodies are found. A full case against each of these views would require separate discussion, but these objections suggest that neither position provides a powerful basis from which to deny the Property View.
Naïve Realism and the Property View
Another
view that rejects the Property view is a simple version of Naïve Realism that
excludes properties from phenomenal character entirely. According to Naïve Realism generally, superstrongly veridical experiences are perceptual
relations to whatever worldly item the subject perceives. According to the
simple version of Naïve Realism, superstrongly veridical
experiences are perceptual relations to objects. Call this the Pure Object
View:
Pure Object View: All superstrongly veridical experiences are perceptual
relations to objects and only to objects.[38]
The Pure Object View is a form of Radical Naïve Realism. A potential proponent of the Pure Object View is Brewer, who writes:
The only alternative to characterizing experience by its representational content is to characterize it as a direct presentation to the subject of certain objects which constitute the way things are for him in enjoying that perceptual experience. Call these the direct objects of experiences: the objects which constitute the subjective character of perceptual experience.[39]
Brewer holds that in cases of illusion, ‘direct objects’ of
experience “have the power to mislead us, in virtue of
their perceptually relevant similarities with other things” (op cit). Since
similarities have to hold in virtue of something, it seems that they would hold
in virtue of properties. And at that point it is hard to see how the resulting
version of Naïve Realism avoids perceived properties.[40]
The main phenomenological objection to Radical Naïve Realism, hence to the Pure Object View arose in defending premise (i) of the Argument from Appearing. In experiences, properties are presented in visual phenomenology, not just objects. To account for the phenomenal character of an experience, Naïve Realism needs some way of specifying which aspects of the perceived object figure in the phenomenal character. Properties (or something in the vicinity, such as centering features or tropes) are well-suited for this job, and it is hard to see how to do without them. Fixing the flaw with the Object View thus leads right back to the idea that properties are presented in experience.
The point that properties are needed to characterize phenomenal character is respected by standard formulations of Naïve Realism, in which the worldly items that partly constitute superstrongly veridical experiences are individuated by both objects and properties.[41] For instance, John Campbell writes:
On a Relational View, the phenomenal character of your experience, as you look around the room, is constituted by the actual layout of the room itself: which particular objects are there, their intrinsic properties, such as colour and shape, and how they are arranged in relation to one another and to you (116, 2002).
Likewise, M.G.F. Martin describes superstrongly veridical experience by invoking both objects and properties:
The Naïve Realist claims that…some of the objects of perception – the concrete individuals, their properties, and the events these partake in – are constituents of the experience (39, 2005).
A similar commitment is made by Kennedy. [42]
To oppose the Content View, these proponents of Naïve Realism need grounds for denying that superstrongly veridical token experiences are accurate with respect to the situation in which they are had. It is hard to see what grounds these might be. According to one line of thought, superstrongly veridical experiences (as standard Naïve Realism construes them) are not accurate, because it is not possible for such experiences to be inaccurate. This line of thought assumes that it makes sense to ascribe accuracy to a state, only if it is possible for instances of that state to be inaccurate. The assumption, however, is wrong. The state of believing that 3+5=8 is never inaccurate. But it doesn’t follow that the belief is not true.[43]
According to a different attempt to deny that superstrongly veridical experiences (as standard Naïve Realism construes them) are accurate, if a state is accurate, then it must be possible to compare the state with the situation of which it is accurate. As standard versions of Naïve Realism construe superstrongly veridical experiences, those experiences contain the relevant situations as constituents. One might conclude from this that no comparison is possible. But there is no metaphysical bar to comparing a state that is partly composed of a situation with the situation of which it is partly composed. (Compare: when a Russellian proposition composed of an object and a property is evaluated with respect to worlds where the object exists, a proposition that is partly composed of an object is compared with a situation containing that very object). The relata are different, even if overlapping.
If the idea of comparison still seems strained, the feeling of strain seems rooted in the redundancy of the perceived situation, which figures on both sides of the comparison. If the state in the alleged comparison is a state of seeing Franco’s sitting down, then in describing the state, we have already described the situation in which the state is had: it is a situation in which Franco is sitting down. Since a situation is not accurate with respect to itself, it may seem as if any comparison covertly targets an aspect of the state that is separable from the things and properties perceived. Since Naïve Realism denies precisely that there are any such separable aspects, such comparison would be illicit.
In response, the Naïve Realist is committed to the idea that perceptual contact with Franco and some of his properties constitutes the subjective character of the experience. When we focus on the subjective character of the experience, comparing it with the situation in which it is had seems to make sense, no matter what metaphysical structure it may have. Franco appears to be sitting down, and we can ask whether things are as they appear. In making this comparison, we don’t have make to any assumptions about the underlying metaphysical structure of the experience. So the idea of comparing a superstrongly veridical experience with the situation in which it is had is not illicit, even if those experiences are structured the way Naïve Realism says they are.
In summary, nothing in the structure of superstrongly veridical experiences as standard Naïve Realism construes them precludes experiences from presenting it as being the case that the object seen has the properties seen, and it is hard to pinpoint what if anything prevents such experiences from having a presentational character. The contrast drawn earlier between experiences and hopes holds independently of whether Naïve Realism is true. If so, it is hard to see the daylight between standard Naïve Realism and the claim that superstrongly veridical experiences are accurate.
Many proponents of Naïve Realism, including Campbell, Brewer, Martin and Mark Johnston, have been vocal critics of the Content View. Given its proximity to the Property View, the most stable way for Naïve Realists to oppose Content View is by embracing the Pure Object View, or some other form of Radical Naïve Realism, which runs afoul of phenomenological considerations. While Radical Naïve Realism can deny the Property View, standard Naïve Realism is hard pressed to avoid it. If so, then it must embrace the Content View as well.
Naïve Realism and the Strong Content View
When proponents of Naïve Realism criticize the idea that experiences have contents, their criticism is best understood as directed at a strong form of the Content View according to which experiences are fundamentally structured as a propositional attitude. One version of this idea is the Strong Content View:
Strong Content View: All visual perceptual experiences consist entirely in the subject’s bearing a propositional attitude toward the contents of her experience.
The Strong Content View is not entailed by either the Content View or the Property View, as neither posits any structure in which properties are presented. It is compatible with both views that experiences (or some subset of them) are fundamentally structured by a perceptual relation, either to external objects (as in Naïve Realism) or to mental objects (as in the classical sense-datum theory). So neither the Content View nor the Property View is committed to the Strong Content View.
Given this difference between the structures posited by Naïve Realism and the Strong Content View, one might think that these views are incompatible. But this claim is an overgeneralization, and the difference in the structures per se is of little philosophical interest. We can see this by considering versions of these views that are clearly at odds, and contrasting them with versions of the views that have close affinities.
Versions the Strong Content View according to which no contents are individuated by perceived objects are clearly at odds with Naïve Realism. And some versions of Naïve Realism are clearly incompatible with the Strong Content View. These include standard Naïve Realism, Radical Naïve Realism, and ‘negative’ disjunctivism, according to which hallucinations consist entirely in a negative epistemic fact, rather than in a mental state with a specific structure.[44]
Other versions of each view, however, bring them closer together. For instance, as McDowell 1996 construes facts, facts are both true propositions, and are also concrete things that can be perceived.[45] There could be a version of standard Naïve Realism that took facts so construed to be constituents of superstrongly veridical experiences. This version of Naïve Realism would clearly be compatible with the Strong Content View.
There are also versions of the Strong Content View that closely resemble standard Naïve Realism. First, according to content disjunctivism, the contents of non-hallucinatory experiences (such as the strongly veridical Lunchtime, or the illusory Fishtank) are individuated by objects that are seen (e.g., the sandwich, the fish, etc), whereas the contents of hallucinations that are indiscriminable from these experiences – such as a hallucination of a sandwich that looks just like your lunch, or of a fishtank – would have contents that are not individuated by any perceived objects.[46] It is also possible to formulate content disjunctivism in a way that individuates the contents of strongly veridical experiences by perceived properties, as well as by perceived objects. The structure of such experiences still differs from the structure posited by Naïve Realism, on the assumption (contra McDowell) that propositions are never also worldly items that can be perceived. But a necessary (and possibly sufficient) condition for entertaining the relevant sort of proposition is to perceive an object and a cluster of its properties. It may seem merely a matter of terminology whether the experience is the entertaining of the object- and property-involving proposition, as per this kind of content disjunctivist, or whether the experience is the perception of objects and properties that gives rise to the entertaining of such a proposition. Whether or not this issue is merely terminological, this sort of content disjunctivism remains closely related to standard Naïve Realism.
Second, there could be a disjunctivist version of the Strong Content View according to which strongly veridical experiences are a variety of factive propositional attitude (such as seeing that p), while other experiences are non-factive propositional attitudes.
In the end, whatever dialectical
status Naïve Realism has in relation to the Strong Content View, it seems clear
that Naïve Realism and the Content View are compatible. Even the forms of
standard Naïve Realism that are incompatible with the Strong Content View are
compatible with the Content View. To reject the Content View while maintaining
Naïve Realism, it is necessary to move away from standard Naïve Realism and
toward the radical form, which I have argued is implausible.
It is easy to get the impression from recent discussions that fundamentally different approaches to perception are exemplified by Naïve Realism on the one hand and the Content View on the other.[47] While the impression of a great divide between these positions is sociologically apt, philosophically it is overdrawn, and it makes a poor guide to the underlying issues. The philosophical divide is not between these approaches per se, but between positions on two questions. The first question concerns whether properties are presented in experience. The Content View stands or falls with the answer to this question. If experiences do not present us with properties (as per Radical Naïve Realism and the raw feel view), then the Content View is false. The second question concerns whether there is any need to individuate experiences by the particular things that the subject of the experience perceives. Naïve Realism stands or falls with the answer to this question. If experiences are not individuated by such objects, then Naïve Realism is false. To the extent that these two questions are independent of each other, Naïve Realism and the Content View are too.
Conclusion: Perception,
Belief and the Strong Content View
Our discussion began with the observation that perception seems to differ markedly from belief. Perception – and with it, perceptual experience - seems to be a special kind of input to belief that allows us to compare our beliefs with the world. I’ve argued that visual perceptual experiences are nonetheless belief-like, in ways brought into focus by the Content View. How, then, can the Content View respect the differences between perceptual experience and belief?
On the assumption that Naïve Realism and the Content View are compatible, one way to respect these differences is to embrace Naïve Realism.[48] According to Naïve Realism, superstrongly veridical experiences have an entirely different structure from beliefs. Whereas beliefs are relations to propositions, superstrongly veridical experiences are perceptual relations to whatever it is that the subject perceives. And of all the things we can perceive, propositions are not among them.[49] Naïve Realism thus offers a clear account of how perceptual experiences (at least the superstrongly veridical experiences) differ from beliefs.
It is also possible to respect the differences between perception and belief while denying Naïve Realism. Even if experiences are belief-like in the ways that are brought into focus by the Content View, they need not be beliefs in order to be fundamentally structured like beliefs. Experiences could be entirely structured as a propositional attitude, yet involve a distinctively perceptual mode of entertaining content – either a mode specific to visual perceptual experiences, or a more generic mode shared by visual perceptual experience and experiences in (some or all) other sensory modalities, but not shared by belief.[50] A distinctively visual perceptual mode of entertaining contents could be thought of as a set of constraints on when a proposition counts as a content of a visual perceptual experience.[51] The constraints would presumably include:
(a) that the proposition is true with respect to a world just in case the experience
is accurate relative to that world;
(b) that the propositions reflect the commitments of the experience;
(c) that they reflect the phenomenal character of the experience; and
(d) that the resulting state is not the kind of state that can and should be adjusted
in response to evidence.
(e) that the proposition is conveyed to the subject
While (a) is also a constraint on belief contents, arguably none of the others are. Constraint (d) does not apply to beliefs, since beliefs are sensitive to evidence: we can and should adjust what we believe to the evidence that bears on them. In contrast, even when experiences may be influenced by non-perceptual states, such as moods, hypotheses, or desires, there does not seem to be such a thing as adjusting your experience in response to evidence, nor any associated norm governing experiences. Constraint (c) also does not apply to beliefs that are standing dispositional states, since these (in contrast to occurrent states) will not have any associated phenomenal character. To the extent that experiences are committal in a different way than beliefs are, (b) is also not a constraint on experiences.[52] Constraint (c) should be taken as encompassing the point that visual experiences are phenomenologically structured by a spatial field – even when they are not perceptual visual experiences, but are rather visual experiences such as ‘seeing stars’ or having phosphenes.[53]
References
Bach, K. 1997, ms, “Searle Against
the World”. Online. Available: http://online.sfsu.edu/~kbach/Searle.html.
Block, N. 1996 Mental Paint and Mental Latex. Philosophical Issues 7: 19-49.
Boring, E. 1929. A History of Experimental Psychology.
Brewer, B. 2006. “Perception and Content” European Journal of Philosophy Volume 14, Issue 2, 165-81.
Byrne, A.2001. “Intentionalism Defended” Philosophical Review 110: 49-90.
---------- 2009. “Experience and Content” Philosophical Quarterly 59. 429-51.
Byrne, A. and Logue, H. 2008. “Either/Or:
Disjunctivism for Dummies.” In Disjunctivism: Perception, Knowledge, Action, eds. F. Macpherson,
A. Haddock, 57-94.
Chalmers, D. 2004. “The Representational Character of
Experience” in The Future for Philosophy,
ed. B. Leiter, 153-81.
----------------- 2006. “Perception and the Fall from
Clarke, A. 2000. A Theory of Sentience.
---------- 2004. “Sensing, Objects and Awareness: Reply to Commentators” Philosophical Psychology 17(4): 553-79.
Dickie,
Dretske, F. 1999. “The Mind’s Awareness of Itself”, Philosophical Studies 95: 103-24.
Dummett, M. 1981. Frege: Philosophy of Language.
Egan, A. 2006. “Appearance Properties?” Nous 40:3: 495-521.
--------------- “Projectivism without Error”, this volume.
Evans, G.1982. The Varieties of Reference.
Farkas, K. 2006. “Indiscriminability and the Sameness of Appearance”, Proceedings of the Aristotelian Society, vol. CVI.
Fish, W., 2009. Perception, Hallucination and Illusion.
---------- 2008. ‘Disjunctivism,
Indistinguishability and the Nature of
Hallucination’. In Disjunctivism: Perception,
Action and Knowledge, eds. F. Macpherson and A.
Haddock, 144-67.
Foster, J. 2000. The Nature of Perception.
Ginet, C., 1975 Knowledge,
Perception and Memory.
Hawthorne, J. and Kovakovich, K. 2006. “Disjunctivism” Proceedings of the Aristotelian Society, Vol 80/1: 145-83.
Heck. R. 2007. "Are There Different Kinds of Content?", in Contemporary Debates in the Philosophy of
Mind, eds. J. Cohen and B. McLaughlin, 117-38.
Kennedy, M, forthcoming. “Heirs of Nothing: Implications of Transparency”. Philosophy and Phenomenological Research.
Koksvik, Ole, ms. "Intuition, Belief and Disposition to Believe".
Leibniz, G.W. 1705/1996, New
Essay on Human Understanding. Trans. and edited by P.
Remnant and J. Bennett.
Lewis, D. 1979 “Attitudes
De Dicto and
De Se”, Philosophical Review 88: 513-43.
Loar,
B. 2005. “Transparent Experience
and the Availability of Qualia”, in Consciousness:
New Philosophical Perspectives, eds Q. Smith and A. Jokic, 77-96.
Locke, J. [1698] 1975. An Essay on Human
Understanding. Ed. P. Nidditch.
Logue, H. 2009. Perceptual Experience: Just Relational, or Representational Too? PhD Diss, MIT.
Martin, M.G.F. 2002. “The
Transparency of Experience”, Mind and Language Vol, 17 No. 4: 376-425.
---------- 2004 “The Limits of Self-Awareness”, Philosophical Studies 120, 37-89.
---------- 2006 “On Being Alienated” in Perceptual Experience, eds
T. Gendler and J. Hawthorne, 354-410.
-------- 1992, “Sight and Touch”, in The Contents of Experience, ed T. Crane,
196-215.
Martin, C.B. 1980. “Substance Substantiated” Australasian Journal of Philosophy 58: 3-10.
McDowell, J. 1996. Mind and World.
McGinn, C. 1989 Mental Content
Pautz, A. 2009. “What Are the Contents of Experience?” Philosophical Quarterly 59: 483-507.
---------2007 “Intentionalism and Perceptual Presence” in Philosophical Perspectives 21: 495-541.
---------- 2010 “Why Explain Visual Experience in Terms of Content?”, this volume.
Peacocke, C. 1992. A Study of Concepts.
Russell, B. 1912/1997. The Problems of Philosophy.
Schellenberg, S., forthcoming.
“Minimalism about Phenomenology”. Philosophy and Phenomenological Research.
Scholl, B. 2002. Objects and Attention.
Sider, T. 2006 “‘Bare Particulars’” Philosophical Perspectives 20: 387-97.
Siegel, S. 2002 Review of Austen Clarke, Sensory Qualities. In Philosophical Review, vol. 111, no 1:
135-38.
------------ 2004 “Indiscriminability
and the Phenomenal”, Philosophical
Studies 120, 91-112.
------------ 2006 “Subject and Object in the Content of Visual Experience” Philosophical Review 115/3: 355-388.
------------ 2008
“The Epistemic Conception of Hallucination”. In Disjunctivism:
Perception, Action and Knowledge, ed. by F. Macpherson
and A. Haddock (
Smith, A. D. 2002 The Problem of
Perception.
Sturgeon, S 2008 “Disjunctivism
about Visual Experience”. In Disjunctivism:
Perception, Action and Knowledge, eds
F. Macpherson and A. Haddock, 112-43.
-------------- 2006 “Reflective Disjunctivism” Proceedings of the Aristotelian Society, vol 80/1: 185-216.
Thompson, B. 2006 “Colour Constancy and Russellian Representationalism”. Australasian Journal of Philosophy 84: 75-94.
--------------- 2009 “Senses for Senses” Australasian Journal of Philosophy. 87/1: 99-117.
--------------- forthcoming. “The Spatial Content of Experience” Philosophy and Phenomenological Research.
Travis, C. 2004 “The Silence of the Senses”, Mind 113: 57-94.
-------------, unpublished. “Comments on Siegel”
Tye, M. 2007 “Intentionalism and the Argument from No Common Content” Philosophical Perspectives. Vol 21/1: 589-613.
Williamson, T. 1990 Identity and Discrimination.
* Starting in 2006,
earlier versions of parts of this paper were presented to audiences at London,
ANU, Glasgow, Stirling, Vermont, Umeaa,
and SPAWN (with comments from Charles Travis), and in graduate seminars at
Cornell, Brown and Harvard/MIT. Thanks to these audiences, and to Nico Silins, Adam Pautz, Daniel Stoljar, Alex
Byrne, Fiona Macpherson, John Morrison, M.G. F.
Martin, Bence Nanay, David
Bain, Kathin Glüer,
Bernhard Nickel, Bill Brewer, Kati Farkas, Joshua Schechter, Chris Hill, Jeff McDonough, Louis deRosset, Matthew Nudds, Matthew Soteriou, and Frank Jackson for further discussion. Many thanks to David Chalmers for numerous conversations about this
material and for much helpful criticism, and to Charles Travis for permission
to quote from his comments.
[1] Other visual experiences are excluded from this category,
such as the experience of ‘seeing stars’ from being hit on the head, or the
experience of having phosphenes. It would also be
natural to use ‘visual perceptual experiences’ more restrictively, to pick out
those experiences that are cases of perceiving, as opposed to hallucinating,
but that use isn’t the one here.
[2] A third constraint will be introduced shortly:
contents are accuracy conditions.
[3] The Content View has the status of an undefended
presupposition in Evans 1982 and Peacocke 1992, as well
in debates about whether phenomenal character supervenes on representational
features, such as having content (e.g., Block 1996 and many of the other papers
bout spectral inversion cited in Chalmers 2004).
[4] Naïve Realism is explained more fully in section 6.
Martin 2002, 2004, Campbell 2002, Johnston 2006, and Brewer 2006 claim that
Naïve Realism is incompatible with the Content View. Brewer argues that
[5] Cf Leibniz: “If our eyes became better equipped or more
penetrating, so that some colors or other qualities disappeared from our view,
others would appear to arise out of them, and we should need a further increase
in acuity to make them disappear too; and since matter is actually divided to
infinity, this process could go on to infinity also." (New Essay
II.xxiii.12/Remmnant-Bennett trans. p. 219)
[6] In contrast to the general notion of acuity at issue
in the here, visual acuity is operationally defined in terms of ability to
identify letters on a Snellen chart. At some levels
of visual acuity, one might not reliably identify a Q (for instance, one might
mistake it for an O), and that would be a misrepresentation at the level of
belief - or if one is guessing, an erroneous guess. The standard notion of visual
acuity does not take a stand on whether the errors that operationally define
such low levels of acuity also occur at the level of experience. For all the
operational definition says, experience may be neutral on the exact
configuration of the letter’s limbs (assuming that an individual letter can be
made out at all), or it may misrepresent them.
[7] Unlike Frege’s notion of
sense, Stümpf and Külpe’s
notion of content does not build in that contents are truth-evaluable.
For discussion, see Boring 1929.
[8] The stronger claim that completely successful
perceptions always lead to knowledge seems false. There may be cases in which
the cube’s greenness plays the right role in producing the experience of seeing
the cube, yet which do not lead to knowledge because the transition from
experience to belief is corrupted.
[9] For a similar case, see
[10] Thanks to Adam Pautz for
suggesting this proposal. In his 2009 Pautz himself
considers the right-hand side of this proposal in a different dialectical
context. He uses it to formulate a disjunctivist
account of accuracy, which he draws on to argue that Naïve Realists can respect
the standard classifications of experiences as accurate or inaccurate.
[11] The proposal faces another objection as well. Since
strong veridicality is defined in terms of indiscriminability,
which is in turn standardly defined in terms of
knowledge (Williamson 1990, Martin 2004 and 2006), it appears that cognitively
unsophisticated creatures cannot count as having strongly veridical
experiences. (More specifically, if a mental state cannot provide a basis for
knowledge in a creature that isn’t capable of knowledge, then it seems that
such creatures can’t have strongly veridical experiences.) This objection is developed in Siegel 2004.
See also Byrne and Logue 2008, Farkas 2006, Hawthorne
and Kovakovich 2006, Pautz
this volume, Sturgeon 2006.
[12] The category should also be taken to include “Spelke-objects”, which are roughly things that can survive
radical changes in kind, but are bounded, capable of continuous motion (and not
of discontinuous motion), and are not parts of other objects. There is debate
among psychologists concerning what principles of individuation govern
perceptual representations of objects. (For discussion, see essays in Scholl
2002 and Dickie ms). A similar question could be
raised specifically about visual phenomenology. Does visual phenomenology ever
attribute properties to Spelke-objects, or is it
neutral on the status of objects as Spelke-objects or
ordinary objects? (The answer will probably
vary between infants and adults). It won’t matter for the Argument from Appearing how this question about visual phenomenology is
answered, as accuracy conditions of the sort described in (iii) can be defined
on any of these construals of objects. But the answer
will bear on which accuracy conditions experiences have.
[13] As bare
particulars are defined here, they are objects without properties. Bare
particulars thus should not be identified with the substrata recognized by
opponents of bundle theory. Substrata are not meant to be (even potentially)
bare particulars in the present sense. The debate between proponents of
substrata and proponents of the bundle theory concerns whether denying the
bundle theory forces you into saying that there are bare particulars in the
present sense. For discussion, see Locke 1689, Martin 1980, Sider 2006.
[14] Clark 2000 suggests that visual phenomenology
presents us only with properties instantiated at locations (‘feature-placing’),
without taking a stand on which objects are instantiating the properties. The
objection that this is phenomenally inadequate in cases of perceived motion can
be found in Siegel 2002. In his 2004,
[15] Dretske 1999. Unlike Clark 2000, Dretske
accepts that properties are sometimes attributed to objects in experience.
[16] Hallucinations that are indiscriminable
from perception will not be cases of seeming to see a bare particular, nor of
seeming to see properties that are not instantiated by anything.
[17] Searle 1983 discusses a closely related category of
mental states, called ‘Bel-states’, of which belief
is supposed to be a paradigm. For Searle the defining feature of Bel-states is their mind-to-world direction of fit, which
is in turn illustrated using the metaphor that it is
the ‘fault’ if the world, not the mind, if the Bel-state
is not satisfied (op cit p. 7). Once we stop relying on the metaphor, the
notion of mind-to-world direction of fit seems best understood as a norm to the
effect that beliefs should be adjusted to fit the evidence, and evidence should
not be gerrymandered to match antecedently formed beliefs. But this
understanding of mind-to-world direction of fit is not useful to delimiting any
class of mental states that includes experiences, since unlike beliefs, experiences are not the kinds of states that can be
supported by evidence.
[18] What about states of visual imagination: do they
present things as being the case? This is a matter of controversy. Imagining
that there is a red cube in my house, or that your uncle is standing on his
head, seems to be as presentational as supposing the same things. There may
also be a simpler kind of imagery that is not presentational, as when I just
imagine a red cube. According to Martin’s Dependency Thesis (Martin 2002),
every case of imagining a red cube is a case of imagining that I am seeing a
red cube. If something like the Dependency thesis is true, then visual imagery
would seem to be thoroughly presentational.
[19] On the plausible assumption that visual phenomenology
presents spatial features such as being nearby the perceiver (considered as the
center), we will count an experience as veridical with respect to a centered
world only when such features are instantiated in it. Can a subject S’s
experience be veridical with respect to a situation in which no subject is
present? For discussion of this question see Siegel 2006, section 5.
[20] According to some philosophers, potential contents of
visual experience are so fundamentally different from potential contents of
beliefs that it is impossible to believe exactly what you experience, and so
the contents of experiences could not be conveyed to the subject in this first
sense. They could, however, be conveyed in a similar sense, if there was a
systematic relationship between experience contents and belief contents.
Providing such an account would need to be done anyway, in order to describe
the differences between beliefs that are closer to the deliverances of
perception and those that are farther removed from it. For discussion, see Heck
2007.
[21] Cf.
Logue ms. Since we bring plenty of standing
representations to bear on perceptual beliefs, one can’t infer from the fact
that one believes that (say) somebody is Franco that the property of being
Franco is presented in visual phenomenology. But what is
at issue here are inferences in the other direction: if a property is presented
in visual phenomenology, then it is natural to attribute that property to
something one sees.
[22] The Ebbinghaus illusion suggests that action is also guided by
visual representations of properties that do not figure in visual
phenomenology. The claim of interest here is not that if action is guided by a
visual representation of some properties, then those properties figure in
visual phenomenology. Rather the claim is that if properties are presented in
visual phenomenology, then they may be fit guide action.
[23] Discovering which properties figure in experience is
difficult, and in many cases introspectively reflecting on an experience is
often little help in deciding whether a property is presented in visual
phenomenology, or farther downstream. For example, introspection alone does not
seem to tell us whether visual phenomenology presents an object as Franco (and
so not as his twin), or as someone with certain facial features that Franco’s
twin could equally share, or as merely as a human-shaped entity that a
non-human alien could equally share.
[24] Someone might press the same objection for the case
of tropes, arguing that it is tropes rather than universals that are presented
in experiences, and that ultimately figure in their
accuracy conditions. The Argument from Appearing could easily be reformulated
to accommodate this position by substituting ‘tropes’ for ‘properties’, except
the consequent of premise (ii) would read ‘things are the way E’s visual
phenomenology presents them only if a cluster of F tropes is
instantiated.’ (For instance, F might be a cluster of red-cubical tropes, where
type of trope is determined by a primitive resemblance relation between
tropes). A trope version of (iii) would look like this:
If
necessarily: things are the way E presents them only if a cluster of F-tropes
is instantiated, then:
there is a set of accuracy conditions for E such that:
the
conditions are satisfied in a world only if there is something that has a
cluster of F-tropes in that world.
.
[25] Premise (i) is also at odds with a view explored by Sturgeon 2006 and
2008 and defended by Fish 2008 and 2009, according to which hallucinations
don’t have any phenomenal character at all. If hallucinations lack phenomenal
character altogether, then (assuming they are nonetheless visual perceptual
experiences), they will be counterexamples to (i).
The idea that hallucinations lack phenomenal character is at odds with the
crudest deliverances of introspection. Fish’s proposal is discussed and
criticized in Siegel 2008.
[26]Premises (ii) and (iii) might be reformulated to
accommodate two-dimensionalist Fregean
views as follows:
2D
Fregean (ii)
If
an experience E presents a cluster of properties F as being instantiated, then:
Necessarily:
things in w are the way E presents them iff something
in w has a cluster of properties that meet the conditions on extension that F
meets in the world where E occurs.
2D
Fregean (iii)
If
necessarily: things in w are the way E presents them iff
something in w has a cluster of properties that meet the conditions on
extension that F meets in the world where E occurs, then:
there is a set of accuracy conditions for E such that:
the
conditions are satisfied in a world iff something in
w has a cluster of properties that meet the conditions on extension that F
meets in the world E occurs.
[27] A potential dialectical difficulty might remain. To
defend a version of the Argument from Appearing that included 2D Fregean (ii) and (iii), what would be needed are reasons to
accept these premises that do not antecedently assume the Content View. The
defenses of Fregean contents given by Chalmers and
Thompson take the Content View for granted. They argue that the Fregean contents are needed to make the right predictions
about the veridicality of experiences involving inversion with respect to color
and spatial features. (Thompson (2006) also argues that his contents are needed
to make the right predictions about color constancy). Perhaps related
considerations about these phenomena could be used to support 2D Fregean (ii) and (iii) in footnote 25 without assuming the
Content View.
[28] Byrne 2009 argues that some uses of “looks F” do
reflect contents of experience, but grants Travis’s semantic objection and goes
on to argue for the Content View on the grounds that it offers the best
explanation of perceptual illusion.
[29]
Although Travis sets out to attack the idea experiences have ‘representational
content’, the characterizations offered of the official target is more
restricted than standard characterizations of the view. First, the target holds that contents can be
‘read off’ of the ways things look. Substituting ‘phenomenal character’ for
‘the ways things look’ would result in the view that for any two experiences
with the same phenomenal character, there is some content that both share. But
some proponents of the idea that experiences have content deny this – such as Block
(1996), who argues that that there are pairs of experiences that have the same
phenomenal character, but differ in their content. Second, substituting ‘known
by introspection alone’ for ‘read off’ results in the claim that introspection
can tell us what contents experiences have. This claim is not entailed by the
Content View. (I argue against it in Siegel 2006). A wider target for Travis (and one that would
make his criticisms more powerful) would look to constraints on contents by
phenomenal character, without endorsing the supervenience
claim, and without importing the assumption that we can discover using
introspection alone what contents experiences have.
[30] Travis op cit, p. 70.
[31] Travis op cit, footnote 12.
[32] Cf Ginet 1975, chapter 5, on objective appearances.
[33] p. 71: “If perception is representational, then for
any perceptual experience, there must be a way things are according to
it…things looking as they do on a given occasion must fix what representational
content experience then has.”
[34] “Things looking (first notion) as they do fixes no
way things should be to be the way the look full stop…There is just too much
things look like…in having the demonstrable looks they do.” p. 78-9. Cf. p. 74: “The
conclusion so far is that on our first notion of looks, looking like
such-and-such cannot contribute to determining how things should be to be the
way the look simpliciter. For so far as it goes,
there is no particular way things should be to be the way they look simpliciter.”
[35] The
second construal of looking that Travis considers is also a notion of public
looks, and so likewise does not directly challenge the Argument from Appearing,
or more generally the idea that the phenomenal character of an experience
constrains its contents. The second notion is expressed by locutions that begin
with “It looks as if…” and take an indicative propositional complement, such as
“It looks as if Pia will sink the putt” (or “It looks
like Pia will sink the putt” – these are Travis’s
examples). Like demonstrable looks, these (putative) facts are also supposed to
be public facts about how things look, rather than facts about mental states:
Travis writes: “It cannot look as if X on this notion where it is perfectly
plain that X is not so.” (76). It might be ‘perfectly plain’ that the sphere is
to the right of a green cube, while nonetheless looking as if it is alone on
the table to someone who is blind in their right visual field. If ‘it looks as
if p’ were reporting a contentful experience, then it could certainly look to a subject S as if X, even if it were
perfectly plain that X were not so – for instance, if S were hallucinating an
airport lounge while standing alone on an empty beach.
[36] ‘Property’ can be construed broadly to include either
universals or tropes or centering features, because none of these options
affects the dialectic surrounding the Argument from Appearing.
[37] For discussion of depth and other spatial properties,
see Foster 2000. Some versions of the classical sense-datum theory that avoid
the Property View can nonetheless embrace the Content View. Nothing in
Russell’s theory rules out that experience of seeing an (external) apple is
accurate, even on the assumption that the experience consists in the perception
of sense-data. One could consistently hold that such an experience is accurate
if it is caused by something that has properties systematically related to the
“primed” properties of the sense-data. The experience would not be accurate
solely in virtue of a match between the properties presented in experience and
the properties instantiated in the subject’s environment, but might naturally
be considered accurate nonetheless.
[38] The Object View could be extended to
non-hallucinatory experiences generally, including illusions. If extended it in
this way, it would become a version of Naïve Realism on which the main division
among experiences is between hallucination on the one
hand, and non-hallucinatory experiences on the other. (Byrne and Logue 2008 call this VI vs. H disjunctivism).
[39] Brewer 2006
[40] Ultimately, Brewer’s theory of illusions assimilates
them to strongly veridical experiences, with the only errors in the picture
located downstream of experiences.
[41] Martin talks about experiences as relations to events
involving properties, rather than objects having properties: e.g., the ball’s
hitting the cube, rather than the fact that the ball it hitting the cube. A
more fine-grained ontology than I will assume here would recognize metaphysical
distinctions between the ball hitting the cube, the ball’s hitting the cube,
and the fact that the ball is hitting the cube, considering the first two
events rather than facts. For the purpose of understanding the relationship
between Naïve Realism and the Content View, it won’t matter if ignore the
ontological differences between these relata, since
they are all individuated at least partly by perceived objects and properties
(the ball, the cube, being hit, being in the process of being hit, etc).
Depending on how the perceived properties are construed, Naïve Realism also
comes in a trope version and a universal version.
[42] Kennedy 2009: "Naïve Realists think of veridical
experience as a relation between subjects and material particulars and their
perceptible properties.”
[43] Of
course experiences as standard Naïve Realism construes them differ structurally
from beliefs. The present point is just that it is not in general true that a
state is assessable for accuracy, only if it is possible for instances of that
state to be false. This is shown by the case of beliefs whose contents are
necessarily true propositions. So no such general point supports that claim
that strongly veridical experiences as Naïve Realism construes them are
assessable for accuracy.
[44] The epistemic conception of hallucination was first
explained and defended by Martin 2004. See also Martin 2006. Pautz (this volume) discusses the distinction between
negative and positive disjunctivism.
[45] McDowell 1996.
[46] For
defenses of content disjunctivism, see Bach ms,
Byrne/Logue 2008, Schellenberg (forthcoming), Tye 2007.
[47] E.g.., Campbell 2002 chapter 6, Martin 2002, Pautz (this volume), Schellenberg
(forthcoming), Logue ms.
[48] Campbell 2002 (chapter 6) has suggested that the
Content View lacks the resources to explain what distinguishes experiences from
propositional attitudes generally (including beliefs), whereas Naïve Realism is
particularly well-suited to do so. His objection is best construed as targeting
the Strong Content View.
[49] As noted a few paragraphs back, McDowell would
disagree with this.
[50] A third account of how visual perceptual experiences
differ from beliefs, besides Naïve Realism and the Strong Content View, appeals
to a difference in kind between their contents, while finding a similarity at
the level of attitude. For instance, there might be a generic committal
attitude, common both to perceptual experiences and to beliefs, with the
difference between perception and belief located exclusively at the level of
contents.
[51] An analogous
point holds for more generic mode of entertaining contents, which could be
thought of constraints on when a proposition counts as a content of a
perceptual experience in any sensory modality.
[52] It would require more discussion to assess whether
experiential commitment is different from the kind of commitment involved in
beliefs. Prima facie it is different. It can visually seem as if two Müller-Lyer lines differ in length, even when one beliefs
that they do not differ. As Ole Koksvik (ms) has
recently pointed out, this does not seem to be a matter of holding
contradictory beliefs, as no irrationality is involved.
[53] For discussion of the spatial field distinctive of
visual experience, see Martin 1992 and Smith 2002.