is color dispositionalism nasty

NENAD MISCEVIC

IS COLOR-DISPOSITIONALISM NASTY

AND UNECOLOGICAL?

ABSTRACT. This article is a brief presentation and defense of response-dispo-sitionalist intentionalism against a family of objections. The view claims that for asurface to have an objective stable color is to have a disposition to cause in

normal observers a response, namely, intentional phenomenal-color experience.The objections, raised recently by M. Johnston, B. Stroud, and by Byrne andHilbert, claim that any dispositionalist view is unfair to the naive perceiver-

thinker, saddles her with massive error and represents her as maladaptated to herenvironment. The paper reconstructs the main line of thought in favor of re-sponse-intentionalism and argues that it is in fact rather charitable and fair to

naıve cognizers, and also avoids a cluster of related objections.

1. INTRODUCTION

Color dispositionalism has been the dominant view of color sinceLocke and Reid until recent times. It has been generalized toresponse-dispositionalism about other secondary and tertiary quali-ties.1 However, in the last decade, it has ceded its titles to its com-petitors, under the joint fire from opposite directions of variousprimary quality views of color on the one side, and of color elimin-ativism on the other. Its two prominent defenders, C. McGinn andM. Johnston, have changed sides around the turn of millennium. Ithink this is an unfortunate situation. In this paper I want to explorethe chances of dispositionalism, and to address some of the recentobjections against it. The pride of place shall be given to the fairnessobjection, alleging that it is not fair to ordinary color-perceivers. Ishall discuss two versions or two prongs of the objection. The first isfocused upon demands of interpreting the ordinary speaker andthinker and her color-talk. Call the philosopher who wishes to derivesubstantial philosophical conclusion from constraints on interpreta-tion ‘‘interpretivist’’. It has been argued by the interpretativists like

Erkenntnis (2007) 66:203–231 � Springer 2007DOI 10.1007/s10670-006-9036-8

Stroud (2000, 2002) that any (response-)dispositionalist theory isuncharitable since it saddles naıve cognizers with massive error andintroduces a gap between seeing and thinking. It has also been arguedby more naturalistic authors, including Johnston (1998), Byrne andHilbert (2007), that it is unacceptable because it saddles the naıveperceiver with ecologically deracinated representations of hersurroundings. The two criticism run very much in parallel, the formeralleging error and the later maladaptation, the twin cognitive andpractical mortal sinns. So is the response-dispositionalist view reallyas nasty and unecological as they allege?

Let me start by admitting an additional motivation for concen-trating upon this line of criticism. Stroud, who is the most prominentinterpretivist raising the objection from the alleged lack of charity,concentrates upon the version of response-dispositionalism that I findmost congenial. The version claims that an objective stable color of asurface is a disposition to produce a specific mental response inperceivers, namely intentional seeing of the color of the surface. Letme call the version ‘‘response-dispositionalist intentionalism’’, RI forshort. Stroud has some good and some bad news for intentionalistviews in general and, in particular for RI. The good news is that it is,in his opinion, the best version of dispositionalism. The bad, indeedvery bad news is that it is untenable, being open to regress argument,to accusation of falsely disjoining the content of color perceptionform content of color thought, and of being uncharitable. Now, themost detailed published answers to Stroud, the ones by Byrne,Boghossian and J. Campbell did not even try to defend the attackedRI view. So, given that RI is my favorite brand of dispositionalism, Ithought I might try a single-handed humanitarian intervention in itsfavor.

In the balance of the Introduction I shall summarize the view to bedefended and briefly place it on the map of related views. In Section 2I shall first set out the central criticism. Then, in Section 3 I shallreconstruct the main line of thought in favor of RI, presenting it as acharitable compromise between two competing radical views. Section4 finally summarizes the main replies to the criticism. Along the way Ishall very briefly mention other related objections, and possible linesof answer, without going into any detail. But I hope the bird’s eyeoverview of the debate within a framework charitable to disposi-tionalism might soften the hearts of some of its enemies, and attractthe yet undecided readers. Let me then first present the componentsof response-dependent or -dispositionalist intentionalism.

NENAD MISCEVIC204

First, dispositionalism. The appeal to disposition that produces aresponse is the common core of the Lockean, response-disposition-alist views of color. The general framework is given by some versionof the following characterization. The concept ‘‘F’’ is response-dependent if two conditions hold:

First, it is necessary that a variant of the following equation holds:x is F iff X tends to elicit such-and-such reaction from such-and-such persons under

specified circumstances.Second, the biconditional just stated is to be read from right to left: x is F in virtue ofeliciting a certain response (call this, with C. Wright, 1992, pp. 108ff, ‘‘the right order

of determination’’).

Consider now the options that the rough characterization leavesopen. The first division concerns the response, Lockean ‘‘idea’’ of red.Is it sensation (the Reidian option), or intentionally presentedproperty? We opt for the later, namely, intentional experience as ofphenomenal color, as against Reidian sensationalism.

The next division concerns the satisfiers of the objective colorconcept. In the case of secondary qualities there are two broad kindsof candidate properties. On the one hand, there are subject relativizeddispositions to elicit a given reaction. To take the example of color, adispositional theory would claim that x is canary yellow if x hasdisposition to elicit a canary-yellow-response. This is the Lockeanvariety of response-dispositionalist construal of color.

On the other hand, there are non-relativized or non-relationalproperties. The most obvious candidates are properties of categor-ical bases, that actually cause the reaction. Usually it is clustersof properties of surfaces, reflectance profiles in the case ofcolor. Now, some primary quality theoreticians of color, mostprominently Jackson, assert, quite reasonably, that it is theinvolvement in producing the color response, that makes these,otherwise disjunctive and gerrymandered property clusters intocolors. Such a claim preserves an important dimension of response-dependence in the account of color. Or, take Broackes’s charac-terization (1992):

color of surface S ¼ thewayW inwhich S changes the

ðphenomenally characterizedÞ light:

The clue is given by appeal to phenomenally characterized light,i.e. light as seen or experienced. It is not simply reflecting more oflonger wavelengths, but reflecting light of the kind that looks purple

IS COLOR-DISPOSITIONALISM NASTY AND UNECOLOGICAL? 205

to observers, that defines purple. The reflectance property W, the wayW in which S changes the light, is ultimately characterized in terms ofhow the light is going to be seen by the subject. I propose to considersuch accounts as weakly response-dispositionalist.

The third division concerns the reading of second clause, con-cerning the order of dependence. The traditional Lockean reading isexplanatory-causal. It is the production of purple-response thatbestows the status of purple upon the relevant surface property(subject-relativized disposition, base property, or the way W). Thegiven disposition (or given base) counts as purple-disposition becauseit produces purple-response. (Further issues concern details of‘‘because’’; is it causal, explanatory, or both; see the next section).Alternatively, one may opt for a much weaker, epistemic reading,proposed by Pettit, Jackson and Smith. It is the production of purple-response that guides us, the cognizers in recognizing (objectively)purple surfaces. The epistemic reading goes naturally with non-relational view of which surface properties are colors.

Here is the proposal, and the view to be defended. To have anobjective color is to have a disposition to cause in normal observers aresponse, namely, intentional experience as of phenomenal color.Start with the usual example of standing color. Lucy is looking at aneggplant at broad daylight. What is the purple color she is seeing? It isthe ‘‘seen’’ property of the eggplant, not an introspected property ofexperience. Joining other content intentionalists, I suggests that thecolor seen is part of intentional content of Lucy’s seeing.

This is the variant of the dispositionalist view to be presented andbriefly defended here. It is a variant of traditional response-depen-dentist view, combined with intentionalism, and to be called here forshort ‘‘response-intentionalism (RI)’’. It specifies the response asintentional, as opposed to merely sensational or qualitative, but itdoes not deny the possibility of there being also a sensational orqualitative element to color experience. It thus combines a moderateintentionalism about subjective color content with a version of thetraditional dispositionalist or response-dispositionalist characteriza-tion of objective color. It is to a large extent conservative, since itfollows the lead of traditional Lockean dispositionalism, but divergesfrom the post-Lockean tradition in its characterization of the sub-jective state involved: while the tradition stresses sensation, RIstresses intentional experience. The view is a version of weak inten-tionalism, since it is not directed against qualia, but rather remainsneutral about them (see the taxonomy of intentionalisms in Byrne,2001a, b, c). RI has also a lot of affinities with the view of Sosa (1990,

NENAD MISCEVIC206

1996) and with early views of Mark Johnston (1992) as well withShoemaker’s (1994, 2003) account of phenomenal properties, butdiffers from the last one in characterization of subjective, phenomenalcolor and in being silent about qualia.

Let me anticipate in a few sentences the rationale of the view.Consider the phenomenology of conscious experience. Lucy sees apurple eggplant, not her own sensations or qualia. She might intro-spect the later, but this is not part of seeing itself. Now, her ‘‘seeing’’might not be altogether veridical. So, in order to have an expressionneutral between ‘‘perceiving’’ and mere ‘‘seeming’’ with which tocharacterize Lucy’s intentional experience of purple, let me borrow aterm from Millikan (2000), and call the intentional act ‘‘visaging’’.Lucy visages the purple surface of the eggplant. But what is thispurple characterizing the surface of the eggplant? To Lucy it lookslike a simple, manifest property. Call this property that is beingintentionally presented to Lucy by (or in) her eggplant experiencephenomenal purple or ‘‘P-purple’’ for short. It is the property oftenimplicitly alluded to when one talks about the property of lookingpurple in a stable and constant way. In general, colors presentthemselves to the naive observer as simple properties of objects. Thisis captured by content-intentionalism which thus preserves the kernelof intuitive, naıve view.

Once we unpack our intuitions about color concepts, colors seemto display curious features: they seem to be tied to vision internally, inthe sense that the only criterion determining what color an item has isthe way the item looks. The way colors present themselves suggeststhey are objective, the internal link with appearances suggest they arenot completely so. Philosophers have then turned to science for theresolution of the tension. Science tells us that one can explain colorphenomena without postulating intrinsic color in objects, over andabove their physical properties, and this much was almost certain atthe time of Newton. Moreover, the recent science informs us that thecrucial features of color perception are to be explained by appeal tothe internal organization of the perceiving apparatus: color categories(dividing the continuous physical range into discontinuous segments),and color opponency (contrasting red to green and blue to yellow) areliterally artifacts of our perceptual and conceptual make-up. The bestphysiological theory available (Hurvich’s) suggests that color canhardly be identified with the surface reflectance, since very manycombinations of reflected light yield the same color. A large numberof surfaces with different reflectance properties can therefore beindistinguishable in color appearance. Any two such surfaces are


called ‘‘metameres’’ of each other, and their indistinguishability as tothe color is called ‘‘metameric matching’’. Although the phenomenonof color constancy grants some measure of objectivity, it is by itselftoo weak to counterbalance the contrary evidence. Finally, the recentcomparative research has uncovered a wealth of variation in animalperception of color, including possibly novel colors and unusualpartitions of color space. It suggests that we might have to relativizecolor to particular animal kinds. The evidence seems to point in thefollowing direction: faced with the opposing pulls in naive view, oneshould weaken the assumption that color is completely objective,sacrifice the assumption that it is a categorical property and retain theassumption that it is internally tied to perception. This suggests arelational view, best captured by the response-dispositional analysis.The closest objective item tracked by P-purple and having to do thesurface of the eggplant is the disposition of the surface to give rise toP-purple intentional states. This is exactly the route RI takes, com-bining intentionalism about subjective color content with a version ofthe traditional response-dispositionalist characterization of objectivecolor. To be purple in scientific, realist sense is to have a dispositionto cause in normal observers a response. The response is intentionalphenomenal-purple experience. Lucy’s intentional content so tospeak spreads phenomenal purple on the surface of objects that havethe right disposition.

The answer thus distinguishes two color properties: first, theintentionally presented phenomenal purple, i.e. the non-real phe-nomenal object of Lucy’s state, second, the scientific and real purple,which consists in the disposition of the surface of the eggplant to giverise to her mental response, i.e. intentional presentation of phenom-enal purple.

physical color C,.

i.e, the disposition normal the surfaceof surfaces, fi observer’s featuring of some objectin normal intentional that lookscircumstances state S phenomenal C

gives rise to (a response)

NENAD MISCEVIC208

2. ERROR AND UPROOTEDNESS: THE TWO-PRONGED CRITICISM

Time for criticisms. Dispositional views face a long list of accusa-tions. It is often accused of crass error (e.g. in Byrne’s statement2001b, p.1: ‘‘Any apparent plausibility this has derives largely, Ithink, from erroneous views of perception’’). As mentioned, I willdiscuss a family of objections that can be represented as a singletwo-pronged criticism. Basically, they claim that RI is unfair to thecognizer, since it is an error theory, alleging massive error in humanbelief system. Stroud, who comes from a more interpretivist tradition,has been earlier accusing dispositionalism of being uncharitable. Inthe color book, he exhews the term itself, but preserves the substanceof accusation: RI is an error theory. In general, dispositionalist viewis as instance of the ‘‘unmasking project’’, and its originator, J. Lockeis one of the unmaskers. He also introduces Locke in a quiteuncharitable way, by immediately pointing to his failure to explainthe correlation between color and physical conditions, a slightlyunfair accusation given that in the next three hundred years nobodyhas managed to bridge the same explanatory gap:

Locke was struck by the fact that whatever physical conditions in fact produce aperception of yellow could have produced something else instead, or nothing at all.

For all that is true of the physical movements alone, ‘‘the Wise Architect’’ couldsimply have ‘‘annexed’’ a different kind of psychological item to those movements, orno psycho-logical item at all. The correlation therefore provides no explanation of

why the kind of perception correlated with certain physical movements is aperception of what it is a perception of. Why is it that what arises from just thesephysical conditions is a perception of yellow? (Stroud, 2000, p. 92).

What is exactly the main sin of the dispositionalist? Stroud allegesthat the unmasking project is generally unfair to human cognizers,and we shall discuss this accusation in detail. But his more specificcriticism concentrates upon the accusation that dispositionalist orresponse-dispositionalist caricatures the basic structure of our cog-nitive apparatus, the relation between perception and thought. Onthis line, response-dispositionalists (including, the defenders of RI)claim that (a) our thought of objects as colored is dispositional incontent (e.g. ‘‘The surface of this eggplant is such-as-to-have-thetendency-to-cause-purple-experiences’’, but (b) perception of objectsas colored is categorical. (e.g. ‘‘I see that the surface of this eggplant ispurple’’). Therefore, the proposition believed (in our example the oneexpressed with in the second brackets) is not the proposition seen (theone within the first brackets). As Stroud puts it, what we think orbelieve to be true in cases of what seem to be propositional seeing is


not really the very same as what we see to be true. His generalconclusion is that dispositionalism is a disappointing explanation:

A dispositional theory, if it were correct, would therefore provide a kind of expla-nation of a necessary connection between our perceptions and beliefs about thecolours of things and the truth of those beliefs. Given the perceptions we do or would

get, and the conditions under which we do or would get them, it would have to betrue that objects are coloured. But the theory would explain that necessary con-nection in the disappointing way that any form of idealism would explain it: by

making the world of coloured things dependent on the possibility of our getting thekinds of perceptions we do. (Stroud, 2000, p. 198)

Stroud is not alone. John Campbell, a color primitivist and a Lockeexpert, classifies Locke as error theorist in his ‘‘ManipulatingColour’’ (Campbell, 2006). Although I fully agree with his stressingthe role of practical experience in the formation and justification ofour naıve color-concept, and with his criticism of reason-based viewsof color statements, I am quite puzzled about his use of the label‘‘error theory’’. He tells us that Locke has criticized the naıve colorconcept, and adds that ‘‘those who have followed Locke in holdingthat there are only the microphysical structures and the tendencies toproduce experiences in us have often also agreed that there is anerror that we naively fall into here: that of supposing that colors arecategorical or intrinsic properties of objects, displayed to us invision.’’ After which, he seems to apply the label ‘‘error theorist’’almost for everyone who does not endorse the full-blooded naıvecolor concept.

What is mere error for more traditional philosophers, is malad-aptation and uprootedness for those interesting in a more ecologicalview. M. Johnston has accused his own former response-disposi-tionalist view of presenting a ‘‘deracinated’’ perceiver confronting acolorless reality.

...the picture of our species which emerges from a generalized projectivism is of ananimal whose basic mental habits, capacities, tendencies are not so many mentaladaptations to reality but rather a repertoire of elaborate errors. The environment of

this animal is so deracinated and the animal’s tendencies are so complex that wecannot appeal to a model of adaptation of minded organism to the antecedentdemands of features of the environment to be grasped and dealt with. We can no

longer see mentation as a type of fittedness to what is there, but only as an accidentalelaboration of an inner life of great complexity, an elaboration not fundamentallyconstrained by a responsiveness to how things are. (Johnston, 1998, p. 4)

I assume that he does not really mean that the environment is‘‘deracinated’’ (uprooted from what or whence?), but that the animalis deracinated from its environment, in the picture offered byprojectivists. Now, Alex Byrne and David R. Hilbert in their ‘‘Color

NENAD MISCEVIC210

Primitivism’’ quote enthusiastically Johnston, and develop his accu-sation about a ‘‘scarcely coherent picture of the organism and itsenvironment’’ threatening the dispositionalist:

Second, (reductive) dispositionalist theories, which identify colors with dispositions

to produce psychological effects. Importantly, these psychological effects must not becharacterized in terms of colors, else the dispositional theory will not be reductive. Soa statement like ,Blueness = the disposition to look blue’ does not – at least without

some reductive gloss on the right hand side – express a version of reductive dispo-sitionalism.... A major problem with dispositionalist theories is that they in effect account for thequalitative nature of color by implausibly locating it in the ‘‘dustbin of the mind’’.

The theorist seems caught between a scarcely coherent picture of the organism andits environment, and gainsaying the nature of color. (Byrne and Hilbert, 2007)

This is the more ecological variant of the accusation for saddlying theperceiver with the massive error. This brings us to the question fromthe title of the paper.

3. NEGOTIATING A CHARITABLE COMPROMISE

Is RI really so unfair to the naıve perceiver? It does attribute a certainerror to her. But, no one is perfect and our everyday experiences andfolk conceptualizations offer no guarantee of being error-free. Theerror in question might be like the folk error of taking ‘‘up’’ and‘‘down’’ as absolute properties of space. It is an error but neitherdramatic, nor irrational. Therefore, in order to judge whether RI is fairto the naıve perceiver-thinker we have to retrace the steps that lead toit, and discuss them briefly keeping the main accusation in our focus.Indeed, the intentionalist response-dispositionalist view has been inrecent literature more often presented by opponents, then by actualproponents, and has thus been often treated unsympathetically fromthe outset. A brief and sympathetic presentation of the main tradi-tional line of thought in favor of response-dispositionalism cantherefore do no harm, in spite of its disappointingly programmaticcharacter. I shall compress the line of though in question, and recon-struct in barest outline what I take to be its gist, relying upon con-siderations that are quite well known and have been discussed in theliterature. The result is a simple (perhaps all too simple) argument forRI, that presents it as a charitable compromise between two extremes,the primitivism plus primary quality view(s) on the positive side andthe eliminativism on the negative. Let me start with the argument. Thetransparency of color-phenomenology secures its initial premise.


1. When a normal subject looks at an eggplant under normalcircumstances, full phenomenal purple (P-purple) is being visaged(intentionally experienced) as being on its surface.

The datum is both obvious and robust. The experience istransparent, and indeed famously so, as stressed by authors fromG.E. Moore to Tye and Burge. In other words, there is no expe-riential reason for Lucy, let alone pressure upon her, to think ofP-purple primarily in terms of effects upon herself. The experiencedcolor hides its constitutive relation to the observer. This cruciallydistinguishes the experience of color from the experience of pain-causing devices, traditionally mentioned in the same context, likeLocke’s manna, a laxative inducing stomach pain. It would beexperienced in a quite different way. So would any device thatproduces pain in the thumb, say a thumbscrew. In these cases,victim’s perceptual apparatus does not ascribe to pain-inflictingitems a phenomenal property corresponding to their pain-produc-ing power. The pain-inflicting power is experienced as a negativeaffordance, but not as actual painfulness of the thumbscrew or thelaxative. In contrast, the naıve view of color is of an objectiveproperty presenting itself in the experience, not of a propertyessentially tied to causing experience. Therefore, the ‘‘naıve Fre-gean content’’ is neither neutral nor self-centered causally charac-terized content, like e.g. ‘‘being disposed to cause P experiences insome kind of perceiver’’. In other words, the naıve beliefs aboutcolor treat the color of our eggplant the same way they treat itsshape. There is no reason obvious to the naıve thinker to treatthem differently.

A serious threat to premise 1 comes, however, from a family ofarguments from inversion purporting to show that the relevantphenomenal or appearance property cannot be color. So, ifP-purple is a color, P-purple is not what is seen by Lucy. This lineof thought has prompted Shomaker to deny that of phenomenalcolor-related properties (appearance properties) are themselvescolors. The discussion of the view would demand a separate paper.So let me just mention that this denial is epistemologicallyimplausible. Our first and most basic contact with color comesfrom acquaintance with these properties, and appeal to disposi-tions, powers and the like is late, derivative and to a large extentscience-based. Once the naıve color is lost, not much is left to buildupon.

NENAD MISCEVIC212

2. Phenomenal purple is not a property of subjective state.

Now, why characterize the color visaged as intentionally presentedcolor of the object, rather than as a color sensation or quale? Why isRI better than its cousin, the traditional sensationalist disposition-alism? It is indeed difficult to find a clear affirmation of the view thatcolor itself is a property of subjective state. I managed to find one inthe characterization of the ‘‘qualia-theory’’ of looks, as presented(but not endorsed) by Pettit:

Looks, as conceptualised in this theory, are almost always credited with three

aspects. First, as already mentioned, the reddish look of an object given in per-ception is a property that can in principle be dissociated from the enabling effectsof seeing something red; for short, it is ,effect-independent’ or, as is sometimes

said, ,intrinsic’ property of the perception (cf. Shoemaker, 1994, p. 22). Second, itis an effect-independent property that is manifest for erceivers, in the sense ofbeing registered as such: perceivers see the object as having the red look – theydon’t just see an object that, unnoticed by them, has a red look – if it looks red

to them. (Pettit, 2003, p. 226)

Now, the quotation suggests that perceivers ‘‘see the object as havingthe red look’’, and at the same time ‘‘the reddish look of an objectgiven in perception’’ can be in principle represented as ‘‘,intrinsic’property of the perception’’. This entails that properties seen as beinghad by objects can be dissociated from the object and represented asintrinsic properties of perception itself. I fail to understand how theredness of a tomato can be seen as a readness of perception. Per-ception itself is not red, on any account of perception, even the mostBerkeleyan one. Shoemaker himself, of course, claims somethingdifferent: that on a qualia theory, as he sees it, there is a quale cor-responding to the presented redness of the tomato, and this is clearlycompatible with our premise 2, implying (if we replace ‘‘purple’’ and‘‘eggplant’’ with ‘‘red’’ and ‘‘tomato’’) that there is a phenomenal redpresented on the surface of the tomato. I think we may thereforeconfidently set aside the strange ‘‘qualia theory’’ adumbrated andsubsequently criticised by Pettit; the wise defenders of qualia wouldagree that full phenomenal purple is not presented to the subjectherself a quality of experience, rather than as a property of object.Boghossian and Velleman (1989), for instance, propose that sensa-tional properties of (subjective) visual field are projected upon theobject, which entails being visaged as being visaged on the surface ofit. In other words, even on the qualia account, the color of eggplant isnot experienced as introspected but as seen. This is the minimal lessonof transparency of experience. Moderate intentionalism only reiter-ates this point, without using it against existence of qualia.2


3. Science suggests that phenomenal purple is not on the surfaceof the eggplant.

There is a panoply of well-known science-derived and -backedreasons in favor of this claim. The main ones concern perceptualvariation and metamerism. Reflectance properties yielding thevisaging of purple are a motley. On the other hand, the full phe-nomenal purple is not presented as a motley, i.e. a disjunctiveproperty. It does not look vague or indeterminate giving rise to adisjunctive judgment (like ‘‘I see something in a fog that is either adog or a cat’’). Neither does it look like a disposition, if anything atall does look so.

Of course, the premise 3 is not as popular nowadays as it used tobe. Given the prominence of physicalist or primary quality accountsof color in recent literature,3 one would need to write at leastsomething like a thin book in order properly do defend 3. I shall notattempt to do it, at least not here. Fortunately, the main interpretivistcriticism for lack of charity comes from Stroud, who accepts 3. So atleast for the purpose of defending RI against the interpretivistaccusation, its advocate does not have to go into a long defense of 3.

So, I shall limit myself to claiming that 3 is at least prima facie themost plausible reaction to the actual heterogenity of physical bases ofany given color. What the premise 3 suggests is that it is best to takethe naıve color-concept for what it is, and accept that nothingphysical and recognizable seems to satisfy it to an acceptable degree. Iam not claiming that the perceptual presentation and the corre-sponding naıve concept are sacrosanct, just suggesting that no simpleand homogeneous physical property of purple-appearing surfacescomes close to being the phenomenal purple. For the rest, I have tojust pass the buck and point to the well-known literature where it isdefended (contemporary classics are Hardin, 1988 and later writings,and Maund, 1995).

We are now ready for the crucial step. Taken together, premises1–3 seem to imply eliminativism in regard to colors, and scientistsand philosophers from Galileo to Hardin and Boghossian andVelleman have taken the corresponding step. But one may wonderwhether this ‘‘unmasking’’ suggestion can be mitigated or avoided.If we cast around for the closest actual property of surfaces thathas at least some crucial properties associated with color, we hitupon causal powers/dispositions. This is what dispositionalisttradition proposes next. Following premise 3, we opt for theintentionalist characterization.

NENAD MISCEVIC214

4. The only unitary, non-disjunctive color-relevant propertycommon to all surfaces that appear P-purple is their dispositionto cause visaging of P-purple.

Let me start from the common ground between dispositionalistsand primary-quality realists (anthropocentric realists as well asJaksonians). We can all agree at least about the following: it is thedisposition to cause the visaging response that picks up objectivecolor. This is the minimal response-dispositionalist component. Next,consider the standard practice of determining reference for partlyerroneous naıve concepts. Take weight and mass, and stay withinClassical Mechanics, for the sake of simplicity.4 The two notions areconfused in the naıve view. The naıve view does not understandsweight as being intrinsic to the material body, not as being relationaland dependent on Earth’s mass. Call the property that wouldcorrespond to naıve concept(ion) i-weight. Nest, there is a naıveconcept(ion) of bulk or mass, often confused with naıve weight by theconcept-possessor herself.

Enters classical mechanics. It tells us that there is a simple, non-relational property which explains the main physical phenomenanaively associated with bulk, the naıve mass. So, one charitably takesnaıve mass (bulk)-concept to refer to classical mass. What about i-weight? In contrast to mass, it is harder to identify. The naıve viewdoes not relate weight to Earth at all and sees as a non-relativeproperty of material bodies. Now, classical mechanics points to aunique, homogenous (same for all bodies) physical relational prop-erty that basically does all that weight is supposed to do, i.e. explains‘‘weight phenomena’’. Therefore, the naıve i-weight concept(ion) isdiscarded: there is no i-weigh. Having a determinate weight is nowidentified with having such and such a mass in Earth’s gravitationalfield. The intrinsic weight is replaced with relational weight, i-weightwith r-weight. Most importantly, the relation to the unifying factor isbuilt into the scientific concept.

In what way is the correction performed charitable? Well, charityminimizes the error of the interpretee, when possible, and aims tomaximize the truth and reference of her beliefs, her rationality andher knowledge, given the circumstances.

First, note that practically nobody claims that being charitable tothe naıve interpretee would demand retaining faith in i-weight, as aprimitive property of tables and cabbages, possibly supervening onr-weight, but being infathomable for science. In this case, beingcharitable does not entail postulating properties beyond the reach of


physics. On the contrary, charity is shown in explaining the naıveerror in a way that completely saves the rationality of the interpretee,maximizes truth-likeness of naıve views, and saves a lot of naıveknowledge. The naıve thinker is obviously not irrational in herignoring the constitutive effect of the mass of Earth on the weight,since she does not have adequate evidence that would be sufficient tolead a normal reasoner to arrive at it. It took Newton’s genius tochange matters. Next, naıve view contains a lot of truths, having todo with relative weight of objects (a live pig weights more than anapple), and since these truths are believed reliably and on goodevidence, the naıve thinker has a lot of knowledge.

All this suggests a similar treatment of color. Color is more likeweight than like mass, but perhaps in even worse shape. At worse,there is no unique, homogenous non-relational physical property thatexplains basic phenomenal color data. At best, there are some highlevel groupings (reflectance triples) that roughly correspond to colordata, but they again involve perceiver relative characterizations.5 Butthe unification is still due to the fact that elements from the groupinglook the same color to us. It’s not that data are sacrosanct. It is thatthey are the starting point, and that there is no account of what holdstogether various purple appearances that does not appeal to specificfeatures of our perceptual apparatus. There is no account of whatgives purple appearances their rich web of relations with other color-appearances that does not appeal to the same specific features.Relation to these features is like the relation of weight to Earth’smass: it is the unifying relation that connects various weight phe-nomena. (If you hold the constant of gravity fast, the mass relationswill mirror the ordinary weight relations. But nobody concludes thatreal weight on Earth is just mass, in abstraction from the gravity.) Soby the standards of normal explanation practice, one should builtrelation to those features, that do a lot of explanatory work, into thesophisticated concept of objective color. And the main unifies is thenetwork of capacities of our actual visual apparatus.

Before concluding the dispositionalist’s reasoning to the bestexplanation, let me mention a more technical point, stressed byStroud and others. It is usually put in terms of conditionals aboutpurple. Consider the counterfactual

If our cognitive makeup were different, ripe eggplants would not be purple on theirsurface.

This counterfactual, if true, would ground response-dependentcharacter of purple in the order of determination: it is out makeup

NENAD MISCEVIC216

that makes ripe eggplants purple. But the counterfactual is verycounterintuitive. So, the response-dispositionalist has to block it, andthe usual way is by tying the color specification to the actual per-ceivers, the way we did. The resulting formulation, i.e. our 5, istherefore rigid. Now, an obvious line of criticism of RI as formulatedis that rigidification (i.e. relativization to actual observers and cir-cumstances) blocks response-dependence, by obscuring the order ofdetermination. It has been argued most forcefully by Stroud. Rigid-ification blocks the use of such counterfactuals, so it threatens toblock the appeal to order of determination, and collapse the primary-secondary quality distinction. This runs counter to the secondaryquality status of color. Paraphrasing Stroud, to relativize the defini-tion of color to how observers and circumstances are right now in theworld, would be to identify purple as a non-dispositional property ofobjects. In his eyes this way of taking it would even be compatiblewith the claim that what causes our perception of purple is theproperty of being purple; it could be because objects are purple that,as things happen in the world right now, they are disposed to producewhat we all recognize as perceptions of purple. That is presumablywhat is true in the case of shape (Stroud, 2000, pp. 125ff).

The answer is that it is not the problematic counterfactual, but thewhole scheme of explanation that accounts for the right order ofdetermination. We explain the property of being purple in terms of,inter alia, contribution of the human cognitive makeup, in the waythat makes its dependence on that makeup obvious. It is the explanation,and not the counterfactual dependence that grounds the subject-to-object order of determination. Such an explanation is not purelycausal, because rigidification blocks simple appeal to causality, but itis not utterly divorced from causal considerations, like mathematicalexplanation in pure mathematics. The way is then open to the mainRI thesis.

5. Being purple in scientific sense is being such as to produce inten-tional states of visaging phenomenal purple in (actual) normalobservers under normal circumstances.

Like in the case of weight, the resulting view is a compromise. Itsteers the middle way between two contrasting sets of more radicaloptions. On the one side there is the real ‘‘unmasking’’ program ofthe eliminativist. It does correspond to Stroud’s portret of philoso-pher driven by science who takes the demands of science too far. Ifully agree with Stroud that the program is uncharitable, refusing to


exploit obvious options that would save some knowledge of the naıvethinker.

On the other side there are non-relational views of color. Primi-tivism is interesting, but fraught with difficulties. What about‘‘anthropocentric realizm’’? Is it really necessary to bring the unifyingelement (producing visaging of P-purple) into the characterization ofcolor? Hilbert has famously tried to keep the relation the definition ofcolor itself. Yes, it is the human-based, anthropocentric criterion thatbind together heterogeneous reflectance properties, but the finalbundle of those is characterizable simply by listing its elements, heclaims. But no-one would proceed in this way in defining everydayweight-property. Imagine a ‘‘geocentric’’ weight-theoretician claim-ing that yes, ordinary weight is a geocentric property in the sense ofEarth’s mass and gravity being its determining factor; nevertheless itis a nor-relational property describable merely in terms of howparticular heavy bodies behave on a balance.

Hilbert and Byrne (2007) offer an interesting line of attack in favorof Hilbert’s color programe as against dispositionalism. They arguethat disunity of physical color-bases for a single hue is compatiblewith unity of perceived hue. In order to do this, they warn againsttaking the naıve unitary view as sacrosanct. Since I agree about thewarning, and disagree about their conclusion, I would like to add acomment. They have spotted a weak point in the account M. Johnstongave in favor of dispositionalism in his (1992). They focus uponassumption he calls ‘‘Revelation: The intrinsic nature of canaryyellow is fully revealed by a standard visual experience as of a canaryyellow thing’’ (Johnston, 1992, p. 138).

Next, they unpack the ‘‘fully revealed’’ part as containing twoclaims. The first ascribes infallibility to, and the second, called bythem Self-Intimation ascribes completeness to positive revelations ofexperience. Now, experience of course does not involve presentingreflectance properties as colors. If the second claim, Self-Intimation istrue, then reflectance properties are out. But, they argue convincinglythat Self-Intimation is too strong. Therefore, the fact that the expe-rienced color is silent about its physicalist credentials poses no realthreat to color physicalism, they conclude. I agree that Self-Intima-tion is too strong, and that relying upon it Johnston has taken amisleading shortcut. But there is no need to appeal to such strongclaims as Revelation or Self-intimation, which misrepresent the naıveview as infallible and complete. One should instead appeal to thenormal practice of assigning reference to problematic naive concepts.It is not that the naıve concept dictates what to include into the

NENAD MISCEVIC218

scientific characterization of its referent. It is rather that the prop-erties that unify the disparate candidates for the would-be referentshould be explicitly involved in the characterization. In the case ofcolor the only unifier is the relation to the perceiver and her experi-ence, as they are in the actual world. And the closest relationalproperty is the disposition. We thus end up with a piece of inferenceto the best explanation, which naturally leads to the response-inten-tionalist thesis. (Of course, I can do here no justice to the subtlety ofthe extant defense of Hilbert’s color programe. I can only apologizefor the brevity and extremely programmatic character of these fewremarks. But I hope to have pointed to elements that make RI appearmore plausible than its recent treatments at the hands of its criticswould suggest).6

Note that the RI thesis leads to accepting colors as real even withinthe Absolute Conception of the world: therein the color is real andrelational.

Before concluding the reconstruction let me just mention the issueof securing a bona fide external(ist) content for color experience. Inorder to do it, the dispositionalist can add to the conclusion 5 thefollowing two conjectures.

6. P-purple tracks objective purple (i.e. the relevant disposition).7. P-purple manifests objective purple.

However, in contrast to mind-independent properties, color isrepresented in a more complicated way. The property signified by itsprimary representation, by the ‘‘immediate response’’, is simply thedisposition of the environment to elicit the response in normalcognizers under normal circumstances. The representation gets con-tent by lawfully manifesting the disposition. The property signified bythe conceptual representation is this very disposition, grasped at oneremove, through the mediation of the experiential response. Thegeneral picture in this case is opposite to the one associated withclassical indication: the head – including the sensory apparatus andsomatic states – picks up widely different physical situations andrepresents them under a unitary heading (no pun intended). Theunification by manifestation is at least as much the work of thecognizer and her evolutionary history, as of the situations perceivedor thought of. This is the initial step. But matters are complicated,and the conjectures open a Pandora box of further issues anddifficulties, which we cannot address here.7


4. COLOR REGAINED

I hope to have already sufficiently indicated that the compromisesought by RI is quite fair to the naıve perceiver. So, before con-cluding the defense, I want to discuss the main objection that wouldmake the compromise a non-starter, and which Stroud incorporateswithin his criticism, connecting it to issues of radical interpretation. Itpresents RI as caricaturing the content of perceiver’s and thinker’srelevant intentional states, and concerns the relation of seeing colorto beliefs about color.8 Stroud alleges that response-dispositionalism(and, by implication, RI) is bound to separate color concepts fromcolor perception (Stroud, 2000, pp. 119, 138). According to him re-sponse-dispositionalists (including the defenders of RI) claim that (a)our thought of objects as colored is dispositional in content (e.g.‘‘The surface of this eggplant is such-as-to-have-the tendency-to-cause-purple-experiences’’, but (b) perception of objects as coloredis categorical. (e.g. ‘‘I see that the surface of this eggplant is purple’’).Therefore, the proposition believed (in our example the one expressedwith in the second brackets) is not the proposition seen (the onewithin the first brackets). As Stroud puts it, what we think or believeto be true in cases of what seem to be propositional seeing is notreally the very same as what we see to be true. The same thing is notboth seen and believed. If this is true, things look really bleak forresponse-intentionalist. The option of claiming that, strictly speaking,there is no such thing as propositional seeing at all, is a non-starter, asStroud rightly observes. And the view that the contents of seeing andbelieving are dramatically different wreaks havoc with our ordinarytreatment of perception. Stroud notes that the object of perceptionand of thought are the same in the case of shape; so how come thatthey are not in the case of color? Is color not visible? On the otherhand, if response-intentionalist tried to avoid the asymmetry byequating the two contents, he would immediately run into troublewith regress (as noted by many, e.g. by Boghossian and Velleman,1989). If being purple is having the disposition to cause experiences oflooking purple, then Lucy’s seeing purple would be unpacked asLucy’s seeing disposition to cause experience of looking like havingdisposition to cause experience of looking like, and so on. So, thechoice is between utter implausibility and a catastrophic regress.

I do not believe RI is committed to seeing-believing asymmetry.The reason is simple. We should distinguish the naıve color conceptfrom the sophisticated color concept. The naıve concept is of course,

NENAD MISCEVIC220

categorical (this is what is wrong with it). So the content of naıvebelief is the same as the content of naıve perception. And there is noregress. The difference comes into play only with the sophisticatedconcept used to elucidate both the referent (and not the conceptualcontent) of the naıve belief and the referent (not phenomenal content)of naıve perception. The alleged asymmetry is not between seeing andbelieving, but between naıve seeing-cum-believing and the sophisti-cated explaining of both, as illustrated in the table:

And this asymmetry is not problematic, but normal and welcome,being an instance of many asymmetries between the manifest and thescientific image. Why else would we need science and inquiry, if allthat is manifest were identical with its deep structure? But whatshould the sophisticated believer believe once confronted with a ripeeggplant in broad daylight? First, that she is ‘‘seeing’’ (in the sense ofhaving visual experience of) phenomenal purple Second, that theproperty underlying the property intentionally experienced is thedisposition to produce experience of precisely this present kind.

It has been further objected by Stroud that accepting the scientific,‘‘unmasking’’ premises, like our 2, leads the theoretician to believethere are no colors; but that the theoretician must be able to ‘‘identifyperceptions as perceptions of this or that color without himselfascribing any color to any physical object’’, and this ‘‘cannot bedone’’ (Stroud, 2002, p. 245; the argument is deployed at length in2000, ch. 7). However, this objection underestimates the possibilitiesof bootstrapping: the unmasking theoretician starts in his own casewith the full panoply of commonsense beliefs, and then proceeds byweakening them, as his theorizing progresses, going from ‘‘this ispurple’’ to ‘‘this looks purple to me’’, to ‘‘this has a disposition tolook P-purple to the normal observer’’, where the content of ‘‘purple’’accordingly changes.9 The criticism is usually phrased so as to suggestthat the problem of identifying the color presented in experience issomehow unique to the response-dispositionalist (-dispositionalist).This is misleading and unfair.10 To see why consider the optionalready mentioned, the dynamical view put forward by J. Broackesand mentioned above:

PERCEPTION THOUGHT

NAIVE a has the P-property a satisfies P-conceptSOPHISTICATED seeing without believing a satisfies D-concept


to be dark blue is not crudely to have a disposition to present a single appearance...;it is to present a variety of appearances in a variety of kinds of lightning, according toa constant pattern. (Broackes, 1997, p. 215)

But how do you characterize light phenomenally? By its effect ofobservers, i.e. by typical experiences they have in relevant circum-stances. Broackes himself has been working for years on less thennormal observers, mostly dichromate’s, and his work has interestingand challenging consequences on both issues: first, who counts asnormal and by what standards, and second, what kind of experiencecounts as being an instance of experiencing purple.

Things are more complicated with the disjunctivist, Jacksonian,variety of realism. It relativizes purple to particular observers andcircumstances. But what makes these observers competent purple-detectors? The question has prompted early contemporary color-realists, like Smart, to attempt eschewing appeal to subjectiveexperience completely. But, whatever criteria a Jacksonian uses topick up a person, say Lucy, as a competent purple-detector, we, theresponse-dispositionalists can re-use to pick her up as a color-normalperson. Realists do not mention observers and normality in theirdefinition of color; this does not mean that they do not have to facethe same problem as response-dispositionalists and, for that matter,as everybody else. In short, every theory of color has to address theissue of how to identify looking purple, and most have to cope withaddressing the normality problem. Both problems are vexing, butthey tell nothing at all against response-dispositionalism in particular(nor against its response-intentionalist variety), being completelygeneral and practically almost everybody’s problems. Since thealleged asymmetry between believing and seeing is Stroud’s mainexample of how dispositionalism (allegedly) caricatures our cognitiveachievements, not much is left from the accusation of being‘‘unmasking’’ in an unfair and uncharitable way.

Charity in interpretation dictates that we do not see folk asreferring to nothing whatsoever when referring to directionsconceptualized in the absolutist, folk way. Rather, they are bestinterpreted as managing to refer to the property that is the closestcousin to the intended one. The point is not just minimizing the error,but also to rationalizing it, making it intelligible. Charity and infer-ence to the best explanation go hand in hand. The traditional dis-positionalist or response-dispositionalist thesis honors both. Itcaptures the fact that the closest actual referent for color conceptsand expressions is the disposition of surfaces to cause the targetintentional states. And it does this stressing the right order of

NENAD MISCEVIC222

determination: what makes a surface purple is its state-producingpower, and not the other way around.

However, RI is not literary an error theory. It alleges partial andvenial error, and this is exactly what we do encounter in the case ofcolor experience. The response-intentionalist uses considerations ofcharity and inference to the best explanation in order to find themiddle way between eliminativism suggested by 2. and 3. and thecommonsense conviction of reality of color. He also takes intoaccount important epistemological considerations about humancapacity to recognize similarities and contrasts of colors.

To summarize, RI succeeds in minimizing and rationalizing theerror of the cognizer, given the suggestions of science. With the visualapparatus they have, humans are presented with a lot of evidence forbelieving in homogenous, manifest color properties. Human cognitivefaculties are often blind to the order of determination, but this is whatis to be expected, and this does not make humans irrational. Barringscience, naıve perceivers have no reason whatsoever to doubt theirevidence, and are perfectly rational in their reasoning. Now whatabout truth and knowledge? Naıve perceivers know a lot of truthsabout phenomenal colors, which they mistake for real, objectiveitems. And, like in the case of weight, they know a lot aboutsimilarities and differences of color of ordinary objects (like ‘‘Blood issimilar in hue to ripe tomatoes’’), phenomenal similarities that mirrorobjective dispositional similarities. So, a lot of naıve candidateknowledge is preserved as bona fide knowledge. These considerationsintroduce an interesting dynamical and historical aspect into thestandards of charity and adaptedness. On the response-disposition-alist view we learn essential new things about the world as we movefrom the manifest picture to the scientific picture. Ascribing such acapacity for in-depth discovery and learning might be more, and notless charitable than picturing the naive cognizer statically immersedin a world of primitive qualities, about which there is nothing more tobe learned. And similarly, human adaptedness to the world mightconsist more in having a capacity to inquire, learn and overturninitially given patterns of beliefs, than in being fixedly and routinelyright about a fixed set of naive assumptions and presuppositions.

We now turn to the maladaptation or uprootedness objection,raised by Johnston, and by Byrne and Hilbert (2007). Remember, itclaims that in dispositionalist’s picture the environment of this animalis so deracinated that we cannot speak of adaptation of mindedorganism to the independently present features of the environment.


The dispositionalist will answer that any features represented atthe most primitive level are very probably secondary qualities. Noticefirst that even if colors and other secondary qualities are not them-selves detected (in the sense discussed), they help us detect other,more robust properties, to which organisms adapt themselves. Thinkof the importance of chemoreception, of smell and taste in simpleranimals – and consider especially those that do not have developedvisual perception. Psychologists speak in this context of the‘‘ubiquitous character of olfactory signals’’. The anatomy–in partic-ular the absence of links from the sense-organ to the neocor-tex–suggests that the sub-system subserving smell and taste inmammals is probably in the evolutionary order among the first andmost primitive distal sensory sub-system in animals. Similarly, interrestrial animals color vision helps detecting a wide range of ob-jects: primates which are equipped with full (s.c. trichromatic) colorvision easily detect orange or yellow fruits in the green foliage, thatare hardly visible to color-blind (dichromatic) animals. So, a tri-chromat is nicely adapted to the orange/foliage difference in surfacestructure, antecedently existing and raising demands to its perceptualsystem, to stay with Johnston’s formulations.

We might assume, with authors like Shepard (1997) andE. Thompson (1995) that the task confronting the visual system is todistinguish and recognize more distal items (objects, surfaces) usinginformation contained in more proximal medium (light, ambient air)in an economical way. Sensibility to color helps segment a visuallypresented scene into surfaces and elements that belong together. Itindicates states of objects, e.g. the ripeness of a fruit. In aquaticanimals primitive color vision serves recognition of objects on con-trasting background: a fish that spots prey from below has to rec-ognize dark object against the bright background, the one that huntsfrom above has the contrary task.

The following hint seems to be useful for an adaptive strategy:mark relevant distal contrasts by more proximal contrasts, and usinga complex set of markers (for both distal items and the medium) thatis still less complex than a full-fledged micro-physical structure ofitems represented! Let me remind the reader of the classical exampleof the hot-cold contrast (discussed in first modern debates onsecondary qualities, most prominently by Descartes and Arnauld in‘‘Meditations’’, and ‘‘Objections and Responses’’). It is a typicalprojected contrast, that finely illustrates the way Mother Naturesolved a simple task from which we can learn about more compli-cated ones. The task is to differentiate objective temperature ranges,

NENAD MISCEVIC224

in relation to the needs of organism. Solution: introduce a contrast inthe representing structure (hot versus cold) that is not to be found inthe items represented. And make the animal respond to the contrastrepresented.

Now, what is the simplest way to make the animal respond? Hereis an economical and ecologically safe way for the designer to take:mark the more distal items by representational markers and presentthem as carrying themselves these markers! In the hot-cold example,present the surface temperature above 80�C as being itself hot, notmerely as being represented by a marker ‘‘hot’’. The same is donewith vision. Take a slight visual difference and exaggerate it into astark contrast. The contrast is projected, not indicated; but it can stillusefully indicate the end of a cliff and the beginning of an abyss. Thisis why primitive visual systems tend to exaggerate in the direction ofcontrast. Color enhances recognisability along the same lines.

If this holds, the markers (i.e. the color items organized withinthe structure of phenomenal color space) would reliably and eco-logically anchor the animal in its environment, and animal ten-dencies would be as complex as needed to cope with the distalenvironment plus the medium of light, but not more than neces-sary. Authors like Roger Shepard detail well-supported conjecturesabout specific adaptations to degrees of freedom of terrestrialillumination including light-changes over daytime and atmosphericcircumstances, and inquire into ‘‘trade-offs between chromatic andspatial resolution’’.11 Similarly, E. Thompson is right in stressingthe variability of functions of color detection, that blocks theattempts at non-dispositional account of the content of colorrepresentations. In short, the ecology of non-primary qualities is apromising area, and the research on biological functions of repre-sentations that stand for such qualities offers a wealth of groundsfor ascribing an ecological point to color representations. RImeshes with such considerations: it present content of theresponse–being appeared to with P-purple–as derived fromresponse’s having biological function to bring the item’s dispositionto manifest itself. Please reread the just quoted criticism byJohnston: ‘‘The environment of this animal is so deracinated andthe animal’s tendencies are so complex that we cannot appeal to amodel of adaptation of minded organism to the antecedentdemands of features of the environment to be grasped and dealtwith.’’ And ask yourself who is being uncharitable, the disposi-tionalist or his critics?


5. CONCLUSION

Let me end by returning to the wider context in which Stroud hasoriginally situated the debate. He is worried not only about colors,but about all sorts of secondary and tertiary qualities. Indeed, andcontrary to these worries, if the brief analysis and defense offered inthis section are on the right track, then secondary qualities offer auseful model for understanding and placement of more complicatedfeatures of the manifest picture of the world. Emotional propertieslike being frightful and lovable, conative properties like beingattractive, repulsive, sexy and the like have been considered obviouscandidates for a response-intentionalist treatment. More problematicare aesthetic and moral properties, which to many people do notappear response-dispositional at all. But color does not appear soeither. If the response-intentionalist account of color is correct, itshows that a major property-kind that does not appear to be subject-dependent at all is profoundly such. It is thus color that can serve asthe model for properties that are not manifestly response-dependent,manifestly subjective at all. A generalized response-intentionalismwould then follow in the footsteps of the traditional Lockean line ofthought about secondary qualities. First, some phenomenal quality(being frightful, being lovable) is being visaged or judged as being abona fide property of some item in the world, material object ofphysical event. But then, from general scientific and naturalisticconsiderations one surmises that the phenomenal quality is not abona fide property of the item. On the other hand, from transparencyof the visaging and other intentional states we philosophers concludethat the phenomenal quality is not a property of that subjective state.Now, the only remaining simple and non-disjunctive property rele-vant to the matter and common to all the items causing states pre-senting the phenomenal quality, is this very disposition to give rise tosuch states. Therefore, by principles of charity and by inference to thebest explanation follows the usual conclusion. Having a quality underconsideration is being such as to cause the response of visaging thecorresponding phenomenal quality in right observers under rightcircumstances. This, if successful, would establish response-inten-tionalism about the candidate non-primary quality.

The prospect of generalization gives some prima facie support torelational treatment of color in general, and hopefully to response-intentionalism in particular. It suggests a charitable compromiseconcerning the whole of manifest picture, and an optimistic view of

NENAD MISCEVIC226

human cognitive effort, that proceeds from more subjective to moresubject-independent view of reality, and at the time manages at leastpartly to explain the former in terms of the later.

ACKNOWLEDGEMENTS

I would like to thank participants in Konstanz color conference, theDubrovnik Mental Phenomena course and my colleagues and friendsin Rijeka and Maribor. Special thanks go to Professor Stroud for hissupport and kindness, and to Ralph Shumacher and Kathrin Gluer.

NOTES

1 See, for example, the papers in the volume edited by Menzies (1998).2 I am leaving aside another standard criticism, often put forward both fromeliminativist and from primary-quality view philosophers concerns problems with

defining (what counts as) ‘‘normal’’, and with the appeal to subjective ‘‘look’’ indetermining what is of what color (e.g. Hardin, 1992).3 With authors like D. Hilbert, F. Jackson and A. Byrne as theirs star defenders.4 Sosa has also used the example of weight many years ago (1990), but he stresses the

relativity of experienced ‘‘heft’’ of an object to persons carrying it. I am moreinterested in the relativity to Earth’s mass.5 For a recent discussion see Zoltan Jakab’s paper (Jakab, MS).6 Let me mention an objection that comes from the nearest theoretical neighbors,who adopt relationalism but deny dispositionalism. Such a relationalist non-dispo-sitionalist can agree with premises 1–4, but then claim that it is the primary qualities

themselves, say reflectances, and not their dispositions that should be equated withobjective color. After all, we claimed that disposition ‘‘gives rise’’ to the visaging orthat it ‘‘produces it’’. But dispositions don’t cause. In a recent paper Johanthan

Cohen (2003, p. 31) briefly appeals to the following conceptual consideration. Itmakes sense to claim that a surface S has disposition to look purple, but it is notreally purple. Call the claim R, for ‘‘really purple’’. Dispositionalism cannot makesense of R, whereas a relational primary quality view can, so the later is to be

preferred to the former. I agree that R makes sense, but would argue that it makessense only because it is left indeterminate, with an open contrast class (e.g. as op-posed to looking purple when properly examined by professionals, or being finkish,

and reacting to being looked at, and the like). It should be replaced with

R*: Here is a surface S* that looks purple to all normal people, including experts,

under all kinds of normal light, always and without exception. And its disposition tolook so is not finkish. Still S* is not really purple.

R* is the proper negation of the dispositionalist account of ‘‘purple’’. Now, I am notsure thatR* doesmake sense.What color is S*? Pseudo-purple?What is themysteriouscomponent that makes it non-purple? The argument is inconclusive at best.


7 I discuss the general psychosemantical background to content-by-manifestation inMiscevic (1997). However, let me add a brief note about complications expected inthe case of color. On the view defended, P-purple experience directly represents the

eggplant surface as being P-purple, and indirectly, from the sophisticated, sciencebased standpoint, as being objectively purple, i.e. having the right disposition. It has,so to speak, two different representational contents. The first is acquired by pro-

jection, the spreading of the P-purple over the surface of the eggplant. The second isindicational: by spreading the phenomenal color the experience indicates the relevantdisposition of the surface. The complication would lead one to predict that a simple

externalist, indicational approach, combined with ordinary intuitions about colormight end in a contradiction. Indeed, Byrne offers a nice summary of the simpleexternalist’s predicament in his 2004 paper. He starts with the inversion issue, whichseems to show that the qualia had by Invert and Vert are different in spite of the

sameness of the external surface. So, qualia seem to be intrinsic. Then we plug in thecontent-externalism and arrive atArgument F: P1. Representational properties are extrinsic. P2. Representationalism

is true (for simplicity, in the simple form (...): QC = the property of being anexperience of something’s looking to have color C). P3. Qualia are intrinsic.The three claims are mutually inconsistent, so representationalist seems to be in

trouble. The present proposal suggests that we modify P1. There is a projectiverepresentational property of presenting a phenomenal color, like P-purple, which isnot ‘‘extrinsic’’, but ‘‘inside-out’’ directed. And there is the extrinsic, disposition-manifesting content. Representationalism is true, but we should distinguish two ways

of representing. There is no inconsistency. The cost of the solution is that its pro-ponent has ultimately to come up with an account of how primitive-looking prop-erties are ‘‘projected’’ onto objects. But such an account is probably needed anyway,

since there is no reason to suppose that the naıve view of the world, or the ‘‘manifestpicture’’ is error-free and projection-free, and there is no reason to suppose that noneof projected properties are being presented to the thinker as primitive.

Now the recent discussion of this point has been a very rich one, and I cannot eventry to do justice to its complexity. Let me just mention that it centers around thephilosophical point that in the imagined case of inverted spectrum both the person

hypothetically like us, Lucy the Vert, and her inverted counterpart, Invert, are rightabout the color of things, say our eggplant. Agreed that color is an external property,we lend into trouble. If Invert, who sees the eggplant as yellowish-green, is rightabout its color, then eggplant has two incompatible colors at the same time. In order

to treat Vert and Invert even-handedly, intentionalist then has to deny that phe-nomenal property is a color at all. The salvaging moves proposed all build a dis-positional or causal characterization into the phenomenal item; either into the

phenomenal property itself (Shoemaker, Egan), or into its presumed mode of pre-sentation (Thompson). Thereby, the naıve, view of color as a categorical bona fideproperty of surfaces is lost, and the phenomenology of the situation caricatured.

I propose that we distinguish the philosophical inclination to even-handedness,inaugurated already by Locke in his initial exposition of inversion where he explainsthat ‘‘simple ideas’’ of color cannot go wrong, from the naıve characterization of thesituation. Naively, the Invert is just plain wrong seeing eggplant as yellowish-green,

and Lucy the Vert is dead right. Even-handedness is not a part of the naıve view ofcolor. Philosophically, they are both wrong in a sense, since P-purple and P-yellowishgreen do not exist. And right in a sense, since the disposition in question can be

NENAD MISCEVIC228

described in a relativized way to accommodate the Invert: the surface of the eggplanthas the disposition to produce P-purple visagings in us (Vert-type creatures), and P-yellowish green ones in them, inverts.

A more detailed discussion should also include the refined variant of the inversionobjection derived from Inverted Earth thought experiment, famously proposed by N.Block.

The projected reference of phenomenal color contents certainly stays fixed on theInverted Earth; the traveler’s purple-visaging states go on referring to P-purple to theend of his days. The rest is open for debate.8 Note that intentionalism helps avoid interesting problems raised by A. Byrne(2000). The first is the obvious regress problem, arising from the fact that purple ismentioned on both sides of the characterization. If ‘‘purple’’ stood for the samedispositional property, the response-dispositionalist would end up claiming that

being purple is producing disposition to look to the perceiver as having a dispositionto produce in her the disposition to look... and so on. Phenomenal color is a fineregress stopper, independently identifiable by experience. The second problem, raised

by Byrne against sensationalist dispositionalism is more interesting. For sensationaldispositionalism, the problem is the one of sensing without believing. Byrne helpfullyintroduces the name of ‘‘natural sign theory’’ for theories that combine following

claims:

Imagine that someone is looking at an eggplant, and sees that it is purple. If we

apply the basic outline of the natural sign theory to the perception of the eggplant’scolor, what is going on? First, a property of the eggplant causes a certain sensibleeffect in the perceiver: a purple sensation. Second, she becomes aware that this

sensation is purple. Third, she forms the belief that the eggplant has the power tocause purple sensations (i.e. she believes that the eggplant is purple.) End of story.

Next, come criticism: the eggplant will look purple to the perceiver, Byrne notes.‘‘But now imagine that our perceiver does not in fact believe that the eggplant ispurple – she suspects, say, that the lighting conditions are abnormal. The eggplantwill still look purple to her. And this cannot be accommodated by the natural sign

theory.’’ (Byrne, 2000, p. 28). In short, if we assume X looks P to O, but O does notbelieve that X is P, then O has the right sensations but no corresponding belief. Byclassical sensationalist theory, X does not look P to O. Therefore classical sensa-

tionalism is false. I agree with Byrne, but add that this is good news for RI.Note that RI presents the relation between seeing and believing in a much moreplausible way than its sensationalist competitor, the natural sign theory. Our per-

ceiver will be seeing (visaging) purple, without believing that the eggplant is purple.So, phenomenal color is much better regress-stopper then the sensation or quale.9 Jackson (1998) has proposed the view that color is a role property with similarity

relations (color circle) determining the role, and with various reflectances as realizerproperties. But the structure of color circle is explained by our cognitive organiza-tion. The similarity structure is due to us, so the ‘‘role property’’ is not mind-inde-pendent in the requisite sense. And the realizer properties realize the role property by

causing experiences with the requisite similarity structure. Once this is accepted, therole account almost turns into a version of dispositionalism.10 On the issue of variation in experience of color compare the debate between

G. Harman, S. Shoemaker and E. Sosa in Villanueva (1996).11 Shepard (1997, p. 327).


REFERENCES

Boghossian, P. A. and J. D. Velleman: 1989, ‘Colour as a Secondary Quality’, Mind

98, 81–103.Broackes, J.: 1992, �The Autonomy of Colour�, in D. Charles and K. Lennon (eds.),Reduction, Explanation and Realism, Oxford University Press, Oxford, pp. 421–65.

Reprinted in A. Byrne and D. Hilbert, 1997, Readings on Color, Vol. 1, ThePhilosophy of Color, MIT Press, Cambridge/MA.

Byrne, A.: 2000, �Colors and Dispositions�, Web, Draft 7/10/00.Byrne, A.: 2001a, ‘Do Colors Look Like Dispositions?’, The Philosophical Quarterly

51, 238–245.Byrne, A.: 2001b, �Colors and Dispositions�, web, Draft 7/10/00.Byrne, A.: 2001c, ‘Intentionalism Defended’, Philosophical Review 110, 199–240.

Byrne, A. and D. Hilbert: 2007, �Color Primitivism�, Erkenntnis 66, DOI 10.1007/10670-006-9028-8.

Campbell, J.: 2006, �Manipulating Colours: Pounding an Almond�, in T. S. Gendler

and J. Hawthorne (eds.), Perceptual Experience, Oxford University Press, NewYork.

Cohen, J.: 2003, ‘Color: A Functionalist Proposal’, Philosophical Studies 113, 1–42.

Hardin, C. L.: 1988, Color for Philosophers, Hackett, Indianapolis.Hardin, C. L.: 1992, ‘The Virtues of Illusion’, Philosophical Studies 68, 371–382.Jackson, F.: 1998, From Metaphysics to Ethics, Oxford University Press, Oxford.Jakab, Z.: MS, �Metameric Surfaces: The Ultimate Case Against Color Physicalism

and Representational Theories of Phenomenal Consciousness�, http://www.neologic.net/rd/chalmers/zoltan.html.

Johnston, M.: 1992, ‘How to Speak of the Colours’, Philosophical Studies 68,

221–263.Johnston, M.: 1998, ‘Are Manifest Qualities Response-Dependent?’, The Monist 81,3–43.

Maund, B.: 1995, Colours, Their Nature and Representation, Cambridge UniversityPress, Cambridge.

Menzies, P. (ed.): 1998, Secondary Qualities Generalized, The Monist 81 (special

issue).Millikan, R. G.: 2000, On Clear and Confused Ideas, Cambridge University Press,Cambridge.

Miscevic, N.: 1997, ‘Secondary and Tertiary Qualities: Semantics and Response-

Dependence’, Pacific Philosophical Quarterly 78, 363–380.Pettit, P.: 2003, ‘Looks as Powers’, Philosophy of Mind, Philosophical Issues 13,221–252.

Shepard, R. N.: 1997, �The Perceptual Organization of Colors: An Adaptation toRegularities of the Terrestrial World?�, in A. Byrne and D. Hilbert (eds.), Readingson Color, Vol. 2, The Science of Color, pp. 311–356.

Shoemaker, S.: 1994, ‘Phenomenal Character’, Nous 28, 21–38.Shoemaker, S.: 2003, ‘Content, Character and Color’, Philosophy of Mind,Philosophical Issues 13, 253–278.

Sosa, E.: 1990, ‘Perception and Reality’, in E. Villanueva (ed.), Information,

Semantics and Epistemology, Basil Blackwell, Oxford.Sosa, E.: 1996, ‘Is Color Psychological or Biological? Or Both?’, Philosophical Issues7 (Perception) 67–74.

NENAD MISCEVIC230

Stroud, B.: 2000, The Quest for Reality : Subjectivism and the Metaphysics of Colour,Oxford University Press, New York.

Stroud, B.: 2002, ‘Reply to Boghossian and Byrne’, Philosophical Studies 78,

239–247.Thompson, B.: MS, �Senses for Senses�, http://faculty.smu.edu/bthompso/sen-ses.html.

Thompson, E.: 1995, Colour Vision, Routledge, London New York.Villanueva E. (ed.): 1996, Philosophical Issues 7 (Perception), Ridgeview, Atascadero,CA.

Department of PhilosophyCentral European University,

Nadar u.9, 1051, Budapest,HungaryE-mail: [email protected]


is color dispositionalism nasty

Documents