Bernhard Nickel · [email protected]
Forthcoming in Philosophical Studies
Intentionalism is the claim that the phenomenological properties of a
perceptual experience supervene on its intentional properties. The paperpresents a counterexample to this claim, one that concerns visual groupingphenomenology. I argue that this example is superior to superficially similarexamples involving grouping phenomenology offered by Peacocke (1983),because the standard intentionalist responses to Peacocke’s examples cannotbe extended to mine. If Intentionalism fails, it is impossible to reduce thephenomenology of an experience to its content.
When we look at the world, we have visual experiences. Two important features ofthese experiences are their phenomenology and their content. The phenomenologyof an experience is what it is like to have it; its content is how the experience rep-resents the world as being. INTENTIONALISM is an influential theory about howthese two features are related. The view says, roughly, that phenomenology super-venes on content, so that any two experiences that differ in their phenomenologydiffer in their content. Refinements will be introduced shortly.
Interest in Intentionalism derives, at least in part, from the role it can play in
a reductive project in the philosophy of mind. We may well be able to explain thecontent of mental states in naturalistically acceptable terms.1 If the phenomenol-
1For example, by giving a causal co-variation account of content, as in Fodor (1994) or Stalnaker
ogy and content of experience are as intimately related as Intentionalism says, wemay also be able to reduce phenomenology to content and thus make it naturalis-tically acceptable. Indeed, this is the use to which Tye (1995) and (2000) want toput Intentionalism. But if Intentionalism fails, that reductive project fails, as well. That is not to say that phenomenology is completely irreducible. For all I say, weare still free to reduce phenomenology to physical or functional properties.
In this paper, I present a counterexample to (a refined version of) INTENTION-
ALISM. My counterexample is not the first; over the last few years, several have
been proposed. One that is often thought to be the strongest appeals to the pos-sibility of inverted spectra. Putative cases of inverted spectra are cases in whichtwo experiences have the same color content but different color phenomenology,hence presenting a counterexample to INTENTIONALISM.2 However, standard ar-guments about inverted spectra rely on substantive and controversial theories ofboth the phenomenology and content of experience—and though these are widelyheld, they are substantive theories.3 Such cases also interact in complex ways withtheories of color properties themselves, for instance, whether the space of color
2Ned Block considers this sort of case decisive (see e.g. Block (1996)); see also Chalmers (2004).
For a putative case of actual spectrum-inversion, see Nida-R¨umelin (1996).
3Which controversial claims they rely on depends on how we spell out the case. There are two
major variants. In the first, we consider two subjects, Vert and Nonvert who are functionally identical. They are looking at a particular red object. The proponent of spectrum inversion submits that theirexperiences have the same content but different phenomenal character. This requires first that thecontent of an experience is determined independently of the phenomenal character of that experience,perhaps by what usually causes experiences of that type. It also requires that we can compare thephenomenal character of experiences across subjects. The first is challenged by, e.g., Siewert (1998),the second by Stalnaker (2003).
In the second variant, we consider a single subject in two different situations. The first situation
is an ordinary one: she is looking at a ripe tomato. She then undergoes some physiological changewhich leaves her spectrum-inverted. For example, she wakes up one morning and says that all ofa sudden, the sky looks yellow, et cetera. Nothing in her environment has changed. Perhaps onthis morning, her experience misrepresents her environment. The proponent of spectrum inversionsubmits that after a period of adjustment, her experience once again veridically represents the colorsof things. The subject once again says that the sky looks blue, grass green, etc. This requires thatthe content of experience can change over time in the way required by the argument, as well as thatthe comparison of qualia in a single subject over time is unproblematic. Again, the first claim is acontroversial claim about how the content of experience is fixed; the second turns out to be false. Thateven the intra-personal case over time is at least problematic is brought out by Shoemaker (1982);see also Harman (1999), Stalnaker (2003). These theorists differ on whether the problems can beovercome. For a very thorough discussion of inverted qualia, see Byrne (Summer 2005).
properties is symmetrical in the way presupposed by the argument.
In contrast to discussions of inverted spectra, my case against INTENTIONAL-
ISM focuses on a very simple counterexample involving no substantive theories of
intentionality. The case I am concerned with involves quite ordinary visual expe-riences all readers will have had—at least by the time they have read this paper. In the course of the paper, I’ll argue that my proposed counterexample is supe-rior to the ones presented by Peacocke (1983) because it is immune to standardintentionalist responses.4
After some preliminaries (section 2), I present my proposed counterexample
(section 3). I then argue that it is indeed a counterexample (section 4).
To say that a state has content is to say that the state represents the world as beingsome way. As such, the state either represents the world veridically or falsidically. It is an assumption of the debate I am addressing that visual experiences havecontent. Moreover, it is an assumption that there is such a thing as the content ofexperience, i.e., that there is a unique way a given experience represents the worldas being.5 By a content property, I shall mean the property of having such-and-suchcontent.
4One case I will not discuss in detail is the duck-rabbit. Two successive experiences of seeing it
as a duck and as a rabbit, respectively, differ in their phenomenology. But there is also an obviousrepresentational difference: in the one case, the experience represents that the lines and dots depict aduck, in the other that they depict a rabbit. Peacocke himself dismisses the example on these grounds(see Peacocke, 1983, p.17).
5Many theorists who otherwise disagree share the assumption that experiences have represen-
tational content, such as Heck (2000), McDowell (1994), Peacocke (1992), Evans (1982), Searle(1983), Stalnaker (1998), Thau (2002), Tye (1995, 2000), Byrne (2001). Some theorists who denythat experiences have content are sense-datum theorists who believe that the sense-data are not them-selves contents, such as Russell (1912/77). For a recent denial, see Travis (2004). 2.2 The Phenomenal Character of Experience
When a subject has a perceptual experience, she is in a phenomenally impressivestate. The feature of the experience that makes it thus impressive is the phenomenalcharacter of the experience. By a phenomenal property I shall mean the propertyof having such-and-such phenomenal character.
In this paper, I will mean by INTENTIONALISM the claim:
[INTENTIONALISM] Necessarily, for a single perceiver, and restricted to visual
perception, if two experiences of this perceiver differ in their phenomenalcharacter, they differ in their content.
As I have formulated it, INTENTIONALISM is a relatively weak claim, and it isdefended by several theorists.6 It is merely the claim that one set of propertiessupervenes on another. Stronger versions are also defended sometimes. There aretwo principal ways of strengthening INTENTIONALISM. First, we can lift the re-striction to a single perceiver and to visual perception, claiming instead that acrossperceivers (and/or across modalities), the phenomenal character of two experiencescannot vary without a difference in their respective contents. Second, we can re-quire a different relation than mere supervenience between the phenomenal andcontent properties of an experience. We might require, for example, that the prop-erty of having phenomenal character C is identical to the property of having thecontent that p, for particular C’s and p’s. The weak version will do for my pur-poses. If I can show that it fails, any stronger version that entails it will also berefuted.
INTENTIONALISM as I have formulated it is compatible with functionalism. In
fact, a theorist who accepts INTENTIONALISM can appeal to functional differencesbetween the deliverances of vision and audition to explain the difference in the
6See for example Byrne (2001), Dretske (1995), Harman (1999), Lycan (1996), Thau (2002), Tye
phenomenal character of visual and auditory experiences that have the same con-tent. She might also appeal to such functional differences to explain the differencebetween seeing and believing, since at least non-occurrent beliefs do not have anyassociated phenomenology.7
At this point, it is important to distinguish between two versions of INTEN-
TIONALISM. Let REDUCTIVE INTENTIONALISM be INTENTIONALISM with the
restriction that one must be able to understand the contents of experience indepen-dently of the phenomenal properties of experience. Let NON-REDUCTIVE INTEN-
TIONALISM be INTENTIONALISM without any such restriction.8 I need to say a bit
more about what NON-REDUCTIVE INTENTIONALISM is and what it is not.
One might hold that one cannot understand what the content properties of
an experience are independently of its phenomenal properties on the followinggrounds. We need to distinguish the contents of experience from other contentfulstates of the agent, such as states of the visual system. A promising line of thoughtholds that a state is an experiential state only if it has some phenomenal property. So in order to identify the contents of experience, we need to have recourse tothe phenomenal properties of experience. On my understanding, holding this viewdoes not bar a theorist from also holding REDUCTIVE INTENTIONALISM.
Rather, a non-reductive intentionalist holds that we cannot understand the con-
tent properties of an experience independently of its phenomenal properties on thefollowing grounds. Even if we take for granted which contentful states are experi-ential states, we still need to appeal to the phenomenal properties of experience toexplain what their contents are. That might be the case, for example, because the
7Thus, Tye (2000, p.62) appeals to a property he calls ‘being poised’ to explain the difference
between a state with the content that p that has an associated phenomenology, and one with the samecontent but without phenomenology. The former, but not the latter state, is “poised” to have an impacton beliefs and desires. And as Tye says, the property of being poised is “essentially a functional roleone” (Tye (2000, p. 62)). Incidentally, this make Tye’s claim mysterious that the property of havinga particular phenomenal character just is the property of having a particular content, since in additionto having a certain representational content, it is also required that the contentful state occupy acertain functional role.
8I follow Chalmers (2004) in drawing this distinction. Paradigmatically, Shoemaker in Shoe-
maker (1994a) and (1994b) defends non-reductive INTENTIONALISM, since on Shoemaker’s view,some of the properties represented in experience are defined in terms of the phenomenal characterthey are associated with.
content of an experience includes that it has such-and-such phenomenal character.
The distinction between REDUCTIVE and NON-REDUCTIVE INTENTIONAL-
ISM is important for two reasons. One concerns the reductive project I mentioned
in the introduction. Only REDUCTIVE INTENTIONALISM can serve in that project. If REDUCTIVE INTENTIONALISM fails, we cannot reduce phenomenal propertiesto content properties.
The other reason the distinction between REDUCTIVE and NON-REDUCTIVE
INTENTIONALISM is important concerns the strategy of this paper. A defender
of NON-REDUCTIVE INTENTIONALISM might try to establish it by making thefollowing two claims:
(1) For any phenomenal character C an experience can have, the contents of that
experience can include the proposition that the experience has C.9
(2) An experience has phenomenal character C if and only if its content includes
the proposition that it has C.
If (1) and (2) are true, then there cannot be a difference in the phenomenal char-acter of an experience without a difference in content, since phenomenal characterdetermines (some) content; hence, a difference in phenomenal character entails adifference in content. Should one wish to defend INTENTIONALISM using (1) and(2), it would still be a substantial claim, since (1) and (2) are substantial assump-tions about experience. But one cannot argue against INTENTIONALISM by coun-terexample, given (1) and (2).10 So my argument cannot, by itself, work against
INTENTIONALISM thus defended. I take it that it at least works against REDUC-
TIVE INTENTIONALISM, and for simplicity, I restrict myself to that claim. In what
follows, I will mean REDUCTIVE INTENTIONALISM by INTENTIONALISM.
9Presumably, not all propositions can be represented in experience, such as the proposition that
the number 2 is even. The restriction to phenomenal characters that an experience can have isintended to ensure that visual experiences need only represent that they have visual phenomenalcharacter, and likewise for the other modalities.
10As Byrne (2001, p. 212) rightly points out, if (1) and (2) are true, even the existence of mental
paint is compatible with NON-REDUCTIVE INTENTIONALISM. For mental paint, see Block (1996).
Consider Figure 1. As you look at it, you can become aware of different groupingsof tiles. Depending on how you look at Figure 1, you can see various sets of squaresas prominent.
Let me number the squares as in Figure 2.
In Figure 1, you can see the squares corresponding to 1, 3, 5, 7, and 9 as prominent,or you can see 2, 4, 6, and 8 as prominent. You may also be able to see othergroupings as relatively more prominent, such as 1, 2, 3, 5, and 8, which form a ‘T’. Consider two successive viewings of Figure 1. Suppose that you see one groupingof tiles during the first viewing, a different grouping during the second. You canhave these different experiences without changing where you look. For instance,you can continue to focus your vision on the center of Figure 1 and still have thedifferent experiences. Let me fix on one particular pair of experiences and namethem:
[E1] You are looking at Figure 1, and you see the corner and center tiles (1, 3, 5,
[E2] You are looking at Figure 1, and you see the four tiles in the middle of each
I will argue that E1 and E2 are a counterexample to INTENTIONALISM: they forma pair of experiences that are had by the same perceiver, are both visual percep-tions, have the same content, but differ in their phenomenology. That E1 and E2
differ in their visual phenomenology seems clear enough. To be sure, ‘visual phe-nomenology’ is a term of art, and perhaps the intentionalist could deny that theintrospectible difference between the experiences is a difference in their phenom-enal properties. For the purposes of this paper, I’ll rest with our intuition that thedifference is in the phenomenal properties.
My argument for the claim that E1 and E2 have the same content consists of twoparts. I will successively argue for these claims:
[PW] The contents of E1 and E2 do not diverge in truth-value in any possible
[SAMENESS] There is no further difference between the contents of E1 and E2.
It follows that E1 and E2 have the same content. Splitting the argument up in thisway allows me to separate two kinds of responses on behalf of INTENTIONALISM:those that deny the first part, and those that deny the second.
Any given content determines a set of possible worlds: the set of worlds in whichthe content is true. This follows simply from the claim that if a state has content, itrepresents the world as being a certain way, and hence can represent it veridicallyor falsidically. Further, everyone will agree that it is sufficient for two contents tobe distinct that they disagree in truth-value with respect to some possible world.11In this section, I will argue that this sufficient condition for the contents of E1 andE2 to be distinct is not met. Now, merely arguing for PW is insufficient to establishthat the contents of E1 and E2 are the same, since we might individuate contentsmore finely than by sets of possible worlds. In that case, we could have a distinctionbetween the contents of the two states without there being any world in which they
11Thus, the arguments I consider are independent of any commitment to possible worlds seman-
tics, either for content-bearing states in general or experiential states in particular.
diverge in truth-value. SAMENESS rules out this possibility, and I will argue for itin the next section.
My strategy for defending PW is to consider some ways of denying it and to
show that none of them are compelling. I follow this strategy, because I think itis intuitively plausible that the two experiences represent the world as being thesame way.12 So it seems best to proceed by considering attempts to articulate sucha difference, and see how they fare.
First Proposal The proponent of INTENTIONALISM might start by pointing tothe feature of grouping itself. On this proposal, E1 represents the squares asgrouped one way, while E2 represents them as grouped another way. Depending onwhat the property of being grouped amounts to, their extensions may not divergein any possible world. So right now, this is just a redescription of the problem.
In response, the proponent of INTENTIONALISM might describe the difference
between the two experiences—their grouping phenomenology—in terms of promi-nence: in E1, certain squares are prominent, in E2 others. This leaves open whetherthe property of being prominent is a dispositional or a relational property. Supposethat being prominent is a dispositional property. That is, being prominent is theproperty of tending to appear prominent to creatures with our particular visual sys-tems when they are in the sort of state that gives rise to the prominence effect. Ifthat is the right way to understand the dispositional reading of being prominent,then the property that the first grouping (1, 3, 5, 7, and 9) has does not diverge inany possible world from the dispositional property the second grouping (2, 4, 6,and 8) has. For all that is required in order for a group of squares to have either
12Incidentally, Wittgenstein seems to hold the same position. In considering the duck-rabbit, he
says that we are torn between two descriptions of what has gone on: on the one hand, you mightdescribe the change as a change in the way the world appears; on the other, you might describe thechange as occurring in the face of an unchanging world. He considers an interlocutor who says: “Butsurely you would say that the picture is altogether different now!”. To this Wittgenstein responds:“But what is different: my impression? my point of view?–Can I say? I describe the alteration likea perception; quite as if the object had altered before my eyes.” (Wittgenstein (1953, Part II, xi))Wittgenstein says that it is as if the object had altered, though he does not endorse that descriptioneither. Rather, he goes on to say, “the expression of a change in aspect is the expression of a newperception and at the same time of the perception’s being unchanged.” (ibid.) On Wittgenstein’sview, it is not clear where to locate the difference between the two experiences.
dispositional property is for it to be arranged as the squares are in Figure 1. Theproperties the intentionalist points to do not diverge in any possible world. Thisproposal is also subject to the criticism I urge against the next one.
Suppose instead that we take the property of being prominent as a relational
property. E1 represents Figure 1 as having the relational property of being suchthat squares 1, 3, 5, 7, and 9 look prominent to the subject of E1; E2 representsFigure 1 as having the relational property of being such that squares 2, 4, 6, and 8look prominent to the subject of the experience. On this view, the two propertiesclearly diverge in some possible world. The actual world is one such world at a timeat which you look at the squares and one grouping (say 1, 3, 5, 7, and 9) seemsprominent. At that time, these five squares have the relational property of seemingprominent to you, while the other squares do not have the relational property ofseeming prominent to you at that time. That is the case because one of the relata isan experience with a certain phenomenology, to wit, the phenomenology of E1.
Whether this is a good defense of INTENTIONALISM depends on whether look-
ing prominent is anything other than an experience’s having a certain kind of phe-nomenal character. If it is not, then we cannot understand the property the expe-rience’s content represents independently of the phenomenal character propertiesof the experience. And in that case, the defender of INTENTIONALISM can onlydefend NON-REDUCTIVE INTENTIONALISM by making this argument. In order todefend REDUCTIVE INTENTIONALISM, one would have to show that the contentat issue can be understood independently of the phenomenal character of experi-ence. Clearly, the same problem plagues the dispositional property proposal fromtwo paragraphs ago. In order to defend REDUCTIVE INTENTIONALISM, the dis-positional property had better not just be that of being disposed to give rise to thisphenomenology.
So this strategy by itself does not work. We might think of the subsequent
proposals as proposals about what the content at issue is.
Second Proposal The defender of INTENTIONALISM could try this. When aset of squares appears grouped together, that is in virtue of their appearing to havesome other feature in common. Such a feature might be that the lines of the squares
that are prominent are thicker, or the lines of the squares that are not prominentslightly blurrier. In that case, the contents of the two experiences diverge. Sup-pose, for example, that the lines of the squares the subject is not attending to aresomewhat fuzzier. Then it is compatible with E1 that the lines making up square 2are fuzzy, while this is not compatible with E2.
The tricky part of this proposal is to implement it in more detail, because the
prominent squares and the non-prominent ones share their lines. For example, inE2, squares 2, 4, 6, and 8 are prominent. If being prominent is a matter of thelines of the square having a certain feature, then the center square 5 should also beprominent, since it shares all of its borders with prominent squares, and so all ofthe lines making up square 5 have the prominent-making feature. So no proposalthat seeks to locate a representational difference between the experience in how thelines are represented will work.
Third Proposal The intentionalist might focus on the squares, rather than thelines, by likening the difference between E1 and E2 to more common cases ofvisual illusion. The most promising version of this approach is to claim that E1and E2 give different illusions of three-dimensionality. In each case, the prominentgrouping of squares visually appears slightly closer to the perceiver than the othersquares. These illusions have different contents.
But this proposal will not do, because of general features of depth perception.
If two objects subtend the same angle in a perceiver’s visual field and one objectappears closer than the other, the apparently closer one also appears smaller.13 Butin the case of the more prominent squares, we do not observe a difference in appar-ent size. The prominent and non-prominent squares have the same apparent size. That strongly suggests that the squares do not appear to be at different distances.
Here we have one reason why my proposed counter-example is more forceful
than ones previously proposed, such as the Necker cube.14 While the Necker cubecan be handled by appeal to differences in the apparent distance from the perceiver,
13See Palmer (1999, pp. 315-8). Palmer offers some striking examples: see especially the figure
Fourth Proposal The intentionalist can try to argue more indirectly. There is alarge body of empirical work on the effects of attention on what a subject sees.15If a subject attends to a particular feature of her visual field, such as a cross dis-played on a computer screen, she may fail to notice other easily visible featuresof that field, such as a small square displayed in the immediate vicinity of thecross. Following Mack and Rock (1998), we can call this phenomenon inatten-tional blindness. Furthermore, it seems apt to describe the difference between E1and E2 by saying that we are attending to a different group of squares. This sug-gests that a subject who attends to one grouping of squares, such as 1, 3, 5, 7, and9, may exhibit a similar bit of inattentional blindness: she may fail, for example,to notice small changes in the squares she is not attending to, though she wouldnotice the same changes were they to occur in the squares she is attending to.
The proponent of INTENTIONALISM might try to parlay these sorts of findings
into an argument that the two experiences differ in their contents. First, she willmake the following observations about what a subject is and is not inattentionallyblind to. When the subject is in E1, she is attending to one set of squares. She willtherefore be inattentionally blind to small changes in the other squares. For thesake of definiteness, let this blindness be that she will not notice if a small blackcircle appears in square 2. When the subject is in E2, she is attending to anotherset of squares, including square 2. Now, she will notice if a small black circleappears in that square. The defender of INTENTIONALISM argues that from theseobservations, it follows that the contents of E1 and E2 are distinct. Since the subjectwould not notice the appearance of a circle in square 2 while she is in E1, the truthof the content of E1 is compatible with the presence of the black circle; but sinceshe would notice the appearance of a circle in square 2 while she is in E2, the truthof the content of E2 is not compatible with the presence of the black circle. Hence,the truth-conditions of the two contents diverge with respect to worlds in whichthere is a black circle in square 2.
However, the claims about content do not follow from the claims about what
15See for example Mack and Rock (1998).
the subject is and is not inattentionally blind to. To make this point, just assume forthe sake of argument that the subject would not notice the appearance of a smallblack circle in square 2 while she is in E1—i.e., while she experiences 1, 3, 5, 7,and 9 as prominent. Suppose that the subject is in E1, the black circle appears insquare 2, and the subject does not notice it. In that case, the subject’s experiencedoes not veridically represent the world. Instead the experience misrepresents theworld as containing nine empty squares. After all, it is not as if the experience issilent on whether there is a black circle in square 2. Therefore, even if the factsabout inattentional blindness are as the proponent of INTENTIONALISM supposesthem to be, E1 is not compatible with the presence of the black circle in square 2. Hence, the content of E1 and E2 do not diverge.
Conclusion I conclude that PW holds. Certainly, logical space has room forfurther proposals. But the ones I have considered seem to be the most compelling. This makes a strong case for PW.
So far, I have argued that the two experiences do not diverge with respect to theirpossible worlds content. This is not enough to show that their contents do notdiverge simpliciter. To make this point, I need to argue that there is no furtherdifference between the contents of E1 and E2, i.e., SAMENESS. In this section,I consider three arguments against this claim and reject them. They address themain alternative conceptions of content extant in the literature: Fregean, Russel-lian, and centered worlds. One upshot of this discussion is that in general, we needto already know that two content-bearing objects have different contents before weargue about how to individuate these contents, i.e., before we argue about Russel-lian or Fregean or possible worlds contents. I shall emphasize this point for thespecial case of experience.
Fregean Contents of Experience Frege famously held that even when two sen-tences have a content that is true in all the same possible worlds, we still need to
distinguish these contents.16 His proposal consisted of two distinguishable parts:a criterion for when two contentful sentences have the same content, and a pro-posal about what sorts of abstract objects we should use to characterize the identityconditions of the contents of sentences.
The first argument I consider tries to generalize Frege’s criterion for when sen-
tences have different contents to experiences.17 It then goes on to argue that, bythis criterion, the contents of E1 and E2 are distinct. For the purposes of this dis-cussion, the second part of Frege’s proposal is irrelevant. Since what matters forthis paper is only whether the contents of E1 and E2 differ, it does not matter howto individuate them if they differ.
Crucially, in order for the criterion for different content-bearing objects to have
distinct contents to be helpful in the current dialectic, it must be such that we canjudge of two experiences whether they have the same content or not, independentlyof having already decided the issue. I shall suggest that the most promising line ofextending the Fregean strategy to experience fails on just this count.
Let me begin with Frege’s original criterion for sentences. Suppose we are
given two sentences, such as “Hesperus is Phosphorus” and “Hesperus is Hespe-rus”. It seems clear that a thinker can accept the latter without accepting the formerwhile being fully rational and understanding the sentences. Whenever this is thecase, we should conclude that the two sentences express different contents. This isFrege’s criterion as applied to sentences. That is, whenever we have a pair of sen-tences that differ in their acceptability to a rational thinker who understands them,the sentences differ in their content. The appeal to a rational thinker is essential. To see this, note that an active sentence and its trivial passive transformation, suchas “John killed Bill” and “Bill was killed by John” can differ in their acceptabilityto some speakers, such as speakers who do not know English very well, or who aredistracted, or apathetic. But that does not show that the two sentences differ in theircontent. The speakers for whom they differ in acceptability are simply irrelevant
16Frege (1997). 17Thus, Peacocke (2001) argues that experiences have Fregean contents, although they are not the
same as the contents of thought. That is why, on Peacocke’s view, it is still true that perception hasa nonconceptual content. I believe that this constitutes a change in position from the one Peacockedefends in Peacocke (1992), which I mention below.
The notion of rational acceptability does not apply straightforwardly to experi-
ences, so the defender of INTENTIONALISM needs to generalize the criterion. Shemight do so as follows:
[FREGE] Whenever two experiences differ in which beliefs it is rational to form
on their basis, they have different contents.
FREGE incorporates the requirement that the thinkers we are considering be ratio-
nal, since not just any beliefs we form on the basis of an experience will count asrelevant.
With FREGE, the defender of INTENTIONALISM can argue that E1 and E2 differ
in which beliefs it is rational to form on their basis, and hence that they differ intheir contents. For example, while E1 makes it rational to believe that the squares1, 3, 5, 7, and 9 are grouped together, E2 makes rational the distinct belief that thesquares 2, 4, 6, and 8 are grouped together. Therefore, the two experiences make itrational to form different beliefs, and hence have different contents. So they cannotbe counterexamples to INTENTIONALISM.
Unfortunately, FREGE is too strong. Suppose we have two experiences that
differ only in their phenomenal character. Clearly, one of these experiences makesit rational to believe that what it is like to have it is like this, while the other makes itrational to believe that what it is like to have it is like that. For this reason, we needto restrict the applicability of FREGE to only certain beliefs we form on the basisof a given experience: those we form based on the content of that experience. Butwe are trying to decide whether the difference between E1 and E2 is a differencein content, and hence appeal to FREGE gets us no further along. Hence, arguingthat we can tell that E1 and E2 have different contents based on the different beliefsthey make it rational to form on their basis does not move the debate along. Weneed a further argument that the difference between E1 and E2 is a difference incontent.19
18This restriction to rational acceptability, rather than acceptability simpliciter, is not controversial.
It is endorsed by such contemporary Fregeans as Peacocke (1992) and Evans (1982).
19An anonymous reviewer suggested the following alternative. Two experiences differ in their
Russellian Contents of Experience The proponent of INTENTIONALISM mightsuggest that the contents of experience are Russellian propositions.20 Russellianpropositions are ordered tuples of objects and properties, rather than senses, andthus impose different identity conditions on contents.21 In the foregoing discus-sion of Fregeanism, I distinguished two aspects of Frege’s position: a criterion forwhen two contentful objects have different contents, and a proposal how best toindividuate contents, given the data the criterion delivers. Defenders of Russellianpropositions provide an alternative to the second part of the proposal, how best tosystematize our data about when two contentful objects have the same or differentcontent. Crucially, they usually do not offer an independent criterion—to replace
FREGE—for when two content-bearing objects have the same content.22
For example, in arguing that experience has Russellian propositions as content,
Peacocke (1992) just assumes that two experiences have different contents, andthen suggests that Russellian propositions offer the best means to capture this dif-ference. His argument concerns the difference between two experiences one mighthave in viewing Figure 3. One can either view it as a diamond, or as a square onits tip.
content just in case a rational thinker can believe that the world is the way it is represented by theone experience but not believe that the world is the way it is represented by the other experience. The problem with this proposal is the same as the one I raise in the text. It is precisely up for grabswhether the experiences make different beliefs about the world rational.
20This is the view Peacocke defends in Peacocke (1992, chp. 3). He calls Russellian propositions
21They need not contain objects. Existentially quantified sentences may also express Russellian
propositions, even though they do not contain any objects.
22If they offer one at all, it usually is no different from FREGE, although defenders of Russellian
propositions may differ on just which pairs of sentences come out as having different contents bythat criterion. I take it that this is one of the issues between Fregeans and Russellians about propernames. The former, but not the latter, claim that one can accept, say, “Hesperus is Hesperus” withoutaccepting “Hesperus is Phosphorus”.
Regarding the difference between these two experiences, Peacocke says this.
Intuitively, the difference between perceiving something as a squareand perceiving it as a (regular) diamond is in part a difference in theway in which symmetries are perceived. When something is perceivedas a diamond, the perceived symmetry is a symmetry about the bi-sectors of its angles. When something is perceived as a square, theperceived symmetry is a symmetry about the bisectors of its sides. [. . . ] how do we capture the nature of the additional content that dis-tinguishes a case in which we not only have veridical perception of asymmetrical region but also have that symmetry perceived?
[. . . ] I suggest that perceptual experience has a second layer of [. . . ]content. [. . . ] These additional contents I call protopropositions.23
As Peacocke says explicitly in this passage, it is an intuitively given premise that thedifference between perceiving figure 3 as a diamond and perceiving it as a squareis a difference in the content of the respective experiences. The only question thatis actually argued is how to capture this difference. So it is just an assumptionof the argument that the difference in experiencing Figure 3 as a diamond andexperiencing it as a square is a difference in content.
That means that considering Russellian contents of experience as opposed to
Fregean ones is a red herring. We do not need different kinds of content, we needdifferent arguments for countenancing a difference in the contents of the two ex-periences E1 and E2. But since Russellians do not have different criteria for whentwo content-bearing objects have different contents, the proponent of INTENTION-
ALISM gains nothing by claiming that the contents of experience are Russellian
Of course, the defender of INTENTIONALISM could say that, just as it is in-
tuitive in the case of Figure 3 that there is a difference in the content of the tworelevant experiences, so it is also intuitive that there is a difference in the contentsof E1 and E2, respectively. Two things should be said about this move. First, I am
23Peacocke (1992, pp. 76-7), emphasis in the original. Tye (2005) endorses the same solution.
suspicious of an unalloyed appeal to intuitions about such semi-technical notionsas representational content.
Second, defending Russellianism about the contents of experience is just half
the battle for the intentionalist. She also has to show that there are suitable proper-ties available that distinguish E1 from E2. As I’ve already said above (section 4.1),some properties will fail the constraints of REDUCTIVE INTENTIONALISM: the in-tentionalist needs to find properties that we can understand suitably independentlyof the phenomenal contrast we are trying to account for.
To bring out the force of this point, it will be useful to distinguish my proposed
counter-example from one that seems quite similar. Consider Figure 4, anotherexample urged by Peacocke (1983).
There is a phenomenal difference between seeing the dots grouped in rows andseeing them as grouped in columns. Proponents of the view that there is a rep-resentational difference between the two experiences have appealed to Russelliancontents of experience.24 When we see the dots grouped as rows, we attribute theproperty to the array of being made up of three rows of three dots each. Whenwe see the dots grouped as columns, we attribute the property of being made upof three columns of three dots each. According to the Russellian the propertiesare distinct—though they are necessarily coextensive—and hence we have differ-ent Russellian propositions. Moreover, these properties can be understood suitablyindependently of the phenomenal contrast to be accounted for.
A similar treatment of my proposed counter-example would look like this. E1
attributes the property of being made up of five squares arranged as 1, 3, 5, 7, and9 are, surrounded by four squares arranged as 2, 4, 6, and 8 are. E2 attributes theproperty of being made up of four squares as 2, 4, 6, and 8 are, surrounded by fivesquares arranged as 1, 3, 5, 7, and 9 are. But the properties attributed here are thevery same properties, and hence even on the Russellian view of propositions, E1and E2 have the same content. Some further property needs to be attributed by E1but not E2. The most natural one is that in E1, the five squares are prominent. Butthat is just our original problem.
A different proposal appeals to different properties. The problem derives from
the fact that even when the five squares 1, 3, 5, 7, and 9 are prominent, the ex-perience still represents the other four non-prominent squares. The last proposalsought to account for that fact by attributing the content that there are four squaressurrounding the five, and that led to disaster. But we can account for the repre-sentation of the four squares differently. According to this proposal, when the fivesquares seem prominent, the experience has the following content: there is a largesquare (the one containing the nine smaller ones) containing five squares arrangedas 1, 3, 5, 7, and 9 are. The four non-prominent squares are simply formed by theoutlines of the large square and the five prominent ones. The content of the secondexperience is then that there is a large square containing four squares arranged as2, 4, 6, and 8 are. These two experiences have contents characterized by differentRussellian propositions.
However, even by the lights of the Russellian, these propositions mischarac-terize the content of the two experiences. Presumably, the general principle forattributing contents to experiences the Russellian deploys is that an experiencerepresents a property just in case that property is prominent. Thus, in the case ofthe columns and rows of dots, the experience represents that there are three rowsjust in case the three rows are prominent. Likewise, the experience represents thatthere are five squares just in case the five squares are prominent. And crucially,the experience represents that there is a large square just in case the large square isprominent. But neither in E1 nor in E2 is the outline of the large square prominentas a whole. That means that the experience does not represent that there is a single
large square. And as I’ve just argued, that is a non-negotiable part of the proposal,since without representing the large square, the experience does not represent thatthere are four squares, either.
That shows that my proposed counter-example is immune from a standard in-
tentionalist response to a superficially similar case. Both might be described asexhibiting grouping phenomena, but they are importantly different. Whereas aRussellian treatment is promising in one case, it is not promising in mine.
Centered Worlds A different argument against SAMENESS begins by observingthat perceptual content is perspectival. It makes a difference to perceptual experi-ence where the subject of the experience is in the world. We need to capture thisfact in the content of perceptual experience, because depending on how exactlythe experience represents the spatial relations between subject and object makes adifference to how the subject acts. One might think that for this reason, there canbe a difference in the content of two experiences without there being a differencein the truth-value of these experiences with respect to any possible world.25
However, this is an unnecessary complication. If there is a difference in content
between E1 and E2 with respect to sets of centered possible worlds, that must bebecause there is a difference with respect to some centered worlds that does notconsist merely in the position of the center. By hypothesis, the two experiencesE1 and E2 do not differ with respect to who has them, with respect to the spatialrelations of the perceiver and the figure, nor with respect to the orientation of thatperceiver. But then, there is a difference in the centered-worlds-content of E1 andE2 only if there also is a difference in the possible-worlds content of E1 and E2. So we cannot show that SAMENESS fails simply by pointing out that the contentsof experience ought to be individuated in terms of centered worlds rather thanuncentered possible worlds.
In making this argument, I tacitly assumed that two centers of a world are
the same if they pick out the same agent, that agent is oriented in the same way,and that agent stands in the same spatial relations to what she sees. The reasonthat these features of the center are the ones that identify the center is that these
are features that are plausibly represented in experience. And as a definitionalmatter about centered-worlds content, the state represents the center as having allof the features that go into individuating that center. Nonetheless, on purely formalgrounds, these identity conditions on centers are only one option among many. Another option important in this discussion is this. Two centers are the same onlyif the agents at the centers have experiences with the same phenomenology. Onthis view, E1 and E2 would have different centered worlds contents, since theywould differ in their center. Intuitively, on this proposal, E1 tells the agent that sheis in a world in which she has a certain phenomenology, and E2 tells her that shehas a different phenomenology.26 This proposal runs afoul of the requirements of
REDUCTIVE INTENTIONALISM, because we pick out the content of experience by
its phenomenal character, and hence cannot understand that content independentlyof its phenomenal character.
So instead of picking out the center of the centered worlds by the phenomenal
character of her experiences, the defender of INTENTIONALISM could pick outthe center by some non-phenomenal property on which the phenomenal propertiessupervene, such as physical or functional properties. This would allow the defenderof INTENTIONALISM to remain with REDUCTIVE INTENTIONALISM and to havea difference in the contents of E1 and E2.
There are two things to say about this proposal. First, it is a substantial claim
about the contents of experience that experiences represent these kinds of phys-ical or functional properties. It certainly seems odd to think that it looks to usas if we have the physical properties that some phenomenal property superveneson. Second, it is not clear that this is a good move for the intentionalist given theproject she is trying to carry out. If she wants to give some reductive account ofphenomenal properties, and she already assumes that phenomenology superveneson physical or functional properties, I cannot quite see what further work INTEN-
TIONALISM does for her. In this case, INTENTIONALISM seems to lose much of
26This is just a different formal representation of the idea I discussed in section 4.1, when I dis-
cussed the proposal that the experiences represent relations between the agent and what she is lookingat. The dialectic here is parallel to the earlier discussion.
Conclusion I conclude that SAMENESS holds, as well. As in the case of PW,logical space has room for further proposals. But the ones I have considered seemto be the most compelling. With PW and SAMENESS in hand, E1 and E2 constitutea counterexample to REDUCTIVE INTENTIONALISM, one that does not succumbto the usual intentionalist responses. Hence, the reductive program that seeks toreduce phenomenal properties to content properties fails as well. Other naturalizingproperties may yet succeed.
Block, N. “Mental Paint and Mental Latex”. In E. Villanueva, ed., PhilosophicalIssues 7: Perception, 19–49 (Atascadero, CA: Ridgeview Press, 1996).
Byrne, A. “Intentionalism Defended”. The Philosophical Review, 110(2), (2001),
—. “Inverted Qualia”. In E. N. Zalta, ed., The Stanford Encyclopedia of Philosophy
(Summer 2005). URL: http://plato.stanford.edu/archives/sum2005/entries/qualia-inverted/
Chalmers, D. J. “The Representational Character of Experience”. In B. Leiter, ed.,
The Future for Philosophy (New York: Oxford UP, 2004).
Dretske, F. Naturalizing the Mind (Cambridge, MA: MIT Press, 1995).
Evans, G. The Varieties of Reference (Oxford: Oxford UP, 1982).
Fodor, J. A. “A Theory of Content, II: The Theory”. In A Theory of Content andother Essays, 89–136 (Cambridge: MIT Press, 1994).
Frege, G. “On Sense and Reference”. In Beaney, ed., The Frege Reader, 151–172
Harman, G. “The Intrinsic Quality of Experience”. In Reasoning, Meaning, andMind, 244–261 (Oxford: Oxford UP, 1999).
Heck, R. G. “Nonconceptual Content and the ”Space of Reasons””. The Philo-sophical Review, 109(4), (2000), 485–523.
Lewis, D. K. “Attitudes De Dicto and De Se”. In Philosophical Papers, Vol. 1,
133–159 (New York: Oxford UP, 1983).
Lycan, W. Consciousness and Experience (Cambridge, MA: MIT Press, 1996).
Mack, A. and I. Rock. Inattentional Blindness (Cambridge: MIT Press, 1998).
McDowell, J. Mind and World (Cambridge, MA: Harvard UP, 1994).
Nida-R¨umelin, M. “Pseudonormal Vision: An actual Case of Qualia Inversion?”
Philosophical Studies, 82(2), (1996), 145–157.
Palmer, S. Vision Science: Photons to Phenomenology (Cambridge, MA: MIT
Peacocke, C. Sense and Content (Oxford: Oxford UP, 1983).
—. A Study of Concepts (Cambridge: MIT Press, 1992).
—. “Does Perception have a nonconceptual Content?” The Journal of Philosophy,
Russell, B. The Problems of Philosophy (London: Oxford UP, 1912/77).
Searle, J. R. Intentionality (Cambridge, UK: Cambridge UP, 1983).
Shoemaker, S. “The Inverted Spectrum”. The Journal of Philosophy, 79(7), (1982),
—. “Phenomenal Character”. Noˆus, 28(1), (1994a), 21–38.
—. “Self-Knowledge and ”Inner Sense”: Lecture III: The Phenomenal Character
of Experience”. Philosophy and Phenomenological Research, 54(2), (1994b),291–314.
Siewert, C. P. The Significance of Consciousness (Princeton, NJ: Princeton UP,
Stalnaker, R. C. Inquiry (Cambridge: MIT Press, 1984).
—. “What might non-conceptual Content be?” In E. Villanueva, ed., PhilosophicalIssues 9: Concepts, 339–352 (Ridgeview: Atascadero, 1998).
—. “Comparing Qualia across Persons”. In Ways a World might be, 219–238
Thau, M. Consciousness and Cognition (Oxford: Oxford UP, 2002).
Travis, C. “The Silence of the Senses”. Mind, 113(449), (2004), 57–94.
Tye, M. Ten Problems of Consciousness (Cambridge, MA: MIT Press, 1995).
—. Consciousness, Color, and Content (Cambridge, MA: MIT Press, 2000).
—. “Nonconceptual Content, Richness, and Fineness of Grain”. In T. Gendler and
J. Hawthorne, eds., Perceptual Experience (Oxford: Oxford UP, 2005).
Wittgenstein, L. Philosophical Investigations, 2nd Edition, Trans. Anscombe (En-
glewood Cliffs, NJ: Prentice Hall, 1953).
MATERIAL SAFETY DATA SHEET I. PRODUCT IDENTIFICATION Chemical Trade Name (as used on label): Chemical Family/Classification: Industrial Nickel Cadmium Storage Battery w/ Pocket Plate(VARTA T, TP and TSP range) Manufacturer's Name/Address: Telephone: For information and emergencies, contact EnerSys'Environmental, Health & Safety Dept. at 610-208-1996 24-Hour Emergency Response
COLLOVELOX Maximum particle refinement is of the utmost importance to your products. It provides optimal efficiency, preservability and physical appearance. These characteristics are readily archived with the Collovelox high speed homogenizing colloïdal mill. Efficiency: The Collovelox effects particle reduction through high pressure, utilizing hydraulic and mechanical cutting, centr