Citation
Defense of Conceptual Analysis and Thought Experiments

Material Information

Title:
Defense of Conceptual Analysis and Thought Experiments
Creator:
LAMMENS, KATHRYN KANUCK ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
A priori knowledge ( jstor )
Cogs ( jstor )
Desert pavements ( jstor )
Hats ( jstor )
Intensionality ( jstor )
Intuition ( jstor )
Ions ( jstor )
Sexually transmitted diseases ( jstor )
Thought experiments ( jstor )
Tin ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Kathryn Kanuck Lammens. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
8/31/2008
Resource Identifier:
436098812 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

DEF ENSE OF CON CEPTUAL ANAL YSI S AND THOU GHT EXPERI MENTS By KATHRYN K ANUCK L AMMENS A DI SSER TATI ON PRESENTED TO THE G RADUATE SCHOOL OF T HE UNI VERSI TY OF FL ORI DA I N PARTI AL FUL FI L L MENT OF T HE REQUI REMENTS FOR THE DE GREE OF DOCTOR OF PHI L OSOPHY UNI VERSI TY OF FL ORI DA 2005

PAGE 2

Copy rig ht 2005 by Kathry n Kanuc k L ammens

PAGE 3

This dissertation is dedicated to my two children, Emma L ouise L ammens and Ma tthew Paul L ammens, w h o m a k e m e s m i l e e v e r y d a y.

PAGE 4

iv ACKNOWL EDGMENTS There are many people whose support, encour ag ement, and w isdom helped make the completion of this projec t possible. I would like to take the oppor tunity to thank them all here . First, I would like to thank Dr. Kirk L udwig , cocha ir of my committee, for seeing me throug h ever y step of this very long pr ocess. His e ncoura g ing w ords and helpful comments on much of what is containe d in this di ssertation have been inva luable. I wo uld lik e to t ha nk Dr . G e ne Witme r, c oc ha ir of my c omm itt e e , n ot o nly fo r h is fee dback a nd sug g estions on drafts of this dissertation, but also for g etting me interested in p hy sic a lis m. I wo uld a lso lik e to t ha nk Dr . G re g Ra y fo r g e tti ng me int e re ste d in thinking about thoug ht experiments during my first semester as a g radua te student. I thank Dr. N ora A lter fr om the depar tment of Ger man and Slavic Studies for se rving on my committee as a n external member. My fa mil y ha s a lso be e n v e ry su pp or tiv e of my stu die s. I wo uld lik e to t ha nk my pa re nts , G e or g e a nd Th e re sa Ka nu c k, wh o r a ise d me to f ini sh e ve ry thi ng I be g in, a nd my sib lin g s, Suza nn e , G e or g e , a nd An dr e w, wh o a re my be st f ri e nd s. I wo uld a lso lik e to thank Ellen Mac car one, who ha s enterta ined many of my questions and who has be en a wonder ful frie nd throug hout my g radua te school y ear s. I would espec ially like to thank Phil L ammens for being the most supportive person in my life by encour ag ing me to do e ve ry thi ng I wa nt t o d o, ho we ve r l on g it t a ke s.

PAGE 5

v TAB L E OF CONTENTS P age A C K N O W L E D G M E N T S ................................................ iv L I S T O F F I G U R E S .................................................... vii A B S T R A C T .......................................................... vii i CHAPTER 1 T H E O R I E S O F C O N C E P T S ............................................ 1 Starting Points and Ter minologica l I ssues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Sketch of The ories of Conce pts: From Fr eg e to Gea ch . . . . . . . . . . . . . . . . . . . . . . . . 9 F r e g e........................................................... 10 W i t t g e n s t e i n ..................................................... 17 R y l e ........................................................... 23 G e a c h .......................................................... 29 Contemporary Theorie s of Concepts: B eale r and Pea cocke . . . . . . . . . . . . . . . . . . . 34 B e a l e r .......................................................... 35 P e a c o c k e ....................................................... 41 T a k i n g S t o c k .................................................... 47 Abstrac t Theories V ersus Use Theorie s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 V a g u e n e s s a n d C o n c e p t s ........................................... 49 O t h e r I s s u e s ..................................................... 50 Contemporary Compli cations: A Posteriori Nec essities . . . . . . . . . . . . . . . . . . . . . . 52 2 F RA ME WORK F OR DE F EN DI NG A T RA DI TI ON AL APP RO AC H . . . . . . . . . . 59 B e g inn ing s o f a Qu a siTr a dit ion a l T he or y of Con c e pts . . . . . . . . . . . . . . . . . . . . . . 60 I ntensions and Two-D imensional Semantics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 I n d i v i d u a t i n g C o n c e p t s ............................................ 64 Con c e ptu a l T ru th, Con c e ptu a l K no wl e dg e , a nd Con c e ptu a l A na ly sis . . . . . . . . . . . 66 C o n ce p t u al T ru t h an d C o n ce p t u al Kn o wl ed ge . . . . . . . . . . . . . . . . . . . . . . . . . . 66 C o n ce p t u al An al y s i s an d C o n ce p t u al Kn o wl ed ge . . . . . . . . . . . . . . . . . . . . . . . . 72 F ra me wo rk fo r T hin kin g Ab ou t T ho ug ht E xpe ri me nts . . . . . . . . . . . . . . . . . . . . . . . 85 Setting Up the F rame work: Canonica l Questions and Scena rio I ntuitions . . . . . 86 Sc ie nti fi c a nd Phi los op hic a l T ho ug ht E xpe ri me nts . . . . . . . . . . . . . . . . . . . . . . . 98 Go od Phi los op hic a l T ho ug ht E xpe ri me nts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

PAGE 6

vi 3 DETAI L S OF THE F RAMEWORK: C ONCEPT POSS ESSI ON AND S O U R C E S O F E R R O R ........................................... 109 Suitable Candidates and Conce pt Possession . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 C o n c e p t P o s s e s s i o n .............................................. 109 I de a l a nd Sui ta ble Ca nd ida te s f or Pa rt ic ipa tio n in Th ou g ht E xpe ri me nts . . . . . 124 Candidates for Conducting Thoug ht Ex periments (Experimenter ) . . . . . . . . . . 127 S o u r c e s o f E r r o r a n d C o r r e c t i o n ....................................... 133 F e w B a d Ph ilo so ph ic a l T ho ug ht E xpe ri me nts . . . . . . . . . . . . . . . . . . . . . . . . . . 133 Ske pti c a l V ie ws Ab ou t th e Ut ili ty of Th ou g ht E xpe ri me nts . . . . . . . . . . . . . . . 138 Proper Per formanc e of Thoug ht Ex periments and Possible Source s of Err or . . 143 Putnam's Twin Ea rth Thoug ht Ex periment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 T w i n E a r t h T h o u g h t E x p e r i m e n t .................................... 146 Var iation of the Twin Ea rth Thoug ht Ex periment . . . . . . . . . . . . . . . . . . . . . . . 150 4 H IS T O R IC A L C H A LLE N G E S T O C O N C E P T U A L A N A LY S IS . . . . . . . . . . . . . 154 P a r a d o x o f A n a l y s i s ................................................. 154 Me no ' s Pa ra do x an d th e Pa ra do x of A na ly sis . . . . . . . . . . . . . . . . . . . . . . . . . . 155 G. E. Mo or e on the Pa ra do x of A na ly sis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 Qu ine a n A tta c ks on Con c e ptu a l A na ly sis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 5 CONTEMPORARY CASE S TUDY: THE Z OMBI E THOUG HT E X P E R I M E N T ................................................. 170 Setting Up the Z ombie Thoug ht Ex periment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 Targ et Scena rio and Canonica l Question of the Z ombie Thoug ht Ex periment . 173 P r o p e r P e r f o r m a n c e R e v i e w e d ...................................... 174 Map of Response s to the Z ombie Thoug ht Ex periment . . . . . . . . . . . . . . . . . . . . . 177 P r o b l e m s w i t h C o n c e p t u a l A n a l y s i s ................................. 181 Problems with the Z ombie Thoug ht Ex periment a nd/or the Concept of Con sc iou sn e ss ............................................... 183 E x c e p t i o n a l i s t A n a l y s e s ........................................... 184 Ascriptive a nd Non-Asc riptive Approa ches . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 Neg lected Phy sicalist Response to the Z ombie Thoug ht Ex periment . . . . . . . . . . . 190 P a r a l l e l C a s e s a n d P a r a l l e l E r r o r s ...................................... 193 V a r i a t i o n o f T w i n E a r t h ........................................... 195 Z o m b i e T h o u g h t E x p e r i m e n t ....................................... 197 C l o s i n g R e m a r k s ................................................... 202 R E F E R E N C E S ....................................................... 204 B I O G R A P H I C A L S K E T C H ............................................. 210

PAGE 7

vii L I ST OF F I GURES Fi gu re p age 1 F r e g e ' s n o t i o n o f a s e n s e .............................................. 13 2 T i l e p r o o f o f t h e P y t h a g o r e a n t h e o r e m ................................... 82 3 A priori proof of the Py thag orea n theore m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 4 A map of r esponses to the zombie thought e x periment . . . . . . . . . . . . . . . . . . . . . . 171

PAGE 8

vii i AB STR AC T Abstrac t of Dissertation Prese nted to the Gra duate School of the Unive rsity of F lorida in Partial Fulf illment of the Requirements for the Deg ree of Doc tor of Philosophy DEF ENSE OF CON CEPTUAL ANAL YSI S AND THOU GHT EXPERI MENTS By Kathry n Kanuc k L ammens Aug ust 2005 Cochairs: Kirk A. L udwig and Dona ld E. Wi tmer Major De partment: Philosophy I n this diss erta tion, I defe nd a tra ditional account of conce ptual analy sis and the use of thoug ht experiments in phil osophy . I then apply this account to a c rucia l case , the zombi e thoug ht experiment. This project has four major contributions. First, it survey s and cla rifies important fe ature s of some historical a nd contempora ry theories of conce pts. Second, it provides a re latively traditional ac count of c on c e pts tha t a c c omm od a te s th e sp e c ia l f e a tur e s o f n a tur a l ki nd c on c e pts . T hir d, it provides a f rame work for thinking about a nd evalua ting thoug ht experiments, which better e nables us to avoid many potential problems in perf orming them. Four th, it offers an applica tion of my analy sis of thought e x periments to a pa rticularly controve rsial case —the infamous zombie thought experiment—and shows how we may be making an e rr or in p e rf or min g tha t e xpe ri me nt. I beg in by g iving a historical fra mework, ske tched in ver y broad stroke s, of theories of conce pts from Fr eg e to Gea ch, culminating in contempora ry theories of

PAGE 9

ix conce pts by Be aler and Peac ocke. I also introduce a contempora ry complication for theories of conce pts, namely , the exist ence of a poster iori nece ssities. I develop a nd ultimately endorse an ac count of c oncepts that a ccommodate s those nece ssities and the specia l fea tures of " natura l kind" conc epts, which none theless re tains many centr al fea tures of tra ditional theories of c oncepts. Once this account of c oncepts is in place , I offe r a f rame work for thinking about and eva luating thoug ht experiments, a fra mework tha t meshes with the ac count of c on c e pts I ha ve e nd or se d. I c on sid e r s ome his tor ic a lly sig nif ic a nt c ha lle ng e s to conce ptual analy sis and, thus, thought experiments. Finally , I apply the appa ratus I have de veloped to the infa mous z ombie thought experiment. I arg ue that the e x periment c an be c riticized using the r esourc es I have develope d. I n particula r, I arg ue that one a ppealing way for the phy sicalist to attack the usual result of the e x periment is by arg uing tha t the participa nt is insufficiently sensitive to the way s in which bac kg round informa tion about the actua l world ca n in fac t be rele vant to the answe r.

PAGE 10

1 CH APT ER 1 THEORI ES OF CONCEPTS I f wha t we do primar ily in philosophy is conce ptual analy sis, then it mi g ht be a rg ue d th a t a ny on e a tte mpt ing to d o p hil os op hy mus t ( a t le a st t a c itl y ) p re su pp os e so me the or y of c on c e pts in d oin g wh a te ve r k ind of ph ilo so ph y he do e s. F or , mi nim a lly , if what we do in philosophy is conce ptual analy sis, we have to make cla ims about them and rely on claims about them, wha tever they turn out to be. I f so, then in orde r to fig ure out what conc eptual ana ly sis is, t he first step is to fig ure out wha t conce pts are, a nd, then, deter mine what we can hope to ge t from ana ly ses of wha tever it is. The nature of the item analy zed will put constraints on what we c an expect as r esults and what the importance of the re sults is . My g oal in Chapter 1 of this diss erta tion is not to provide an e x haustive ac count of all of the dif fer ent theorie s of conc epts. Rather, it is to give a historical f rame work, sketche d in very broad stroke s, of theorie s of conc epts from F reg e (1882, 1891, 1892a , 1892b, 1892c, 1893, 1918) to Ge ach ( 1956) and the n to introduce c ontemporar y theories of conc epts by Be aler (1987, 1992, 1998b, 2002a) and Peac ocke ( 1989, 1992). The purpose of Chapter 1 is not to promote any one theor y of conc epts in particular ; instead, it is t o prese nt a historical fr amewor k for a theory of conc epts so that we c an better fig ure out what we a re doing when we claim to be doing conce ptual analy sis and what we can expect from doing conce ptual analy sis.

PAGE 11

2 Historically , in particular with Fre g e (1882, 1891, 1892a , 1892b, 1892c, 1893, 19 18 ), the on tol og y of c on c e pts pla y s a sig nif ic a nt r ole in t he the or y of c on c e pts , b ut i t has re cede d in contempora ry theories. Conseque ntly , I will spend more time discussing issues that are both historically important and play a role in contempora ry theories of c on c e pts , in pa rt ic ula r t he e pis te mic iss ue of c on c e pt p os se ssi on , o r, in F re g e a n te rm s, g ra sp ing a se ns e , a nd the se ma nti c iss ue of de te rm ina te ne ss ( i.e ., the iss ue of wh e the r i t is alway s determinate whether a conc ept applies or f ails to apply ). L ater in this chapter, I will introduce complica tions presented by a posterior i nece ssities for the theory of conc epts. I will arg ue that if we acc ept the standa rd ac count of a po ste ri or i ne c e ssi tie s e nd or se d b y Kr ipk e (1 98 0) a nd Put na m ( 19 75 ), we ha ve to individuate conce pts differe ntly than we ma y first have thought a nd be ca ref ul about what we require for c oncept possession, since if we w ant to maintain a more traditional the or y of c on c e pts , w e sh ou ld n ot w a nt t o r e qu ir e on e to h a ve e mpi ri c a l kn ow le dg e to possess a conc ept. I n Chapter 2 a nd Chapter 3 of this diss erta tion, I will ex plain in more detail some aspe cts of a the ory of conc epts that I will use in the remainde r of the dissertation. Starting P oints and Term inologi cal Issues Although the primary focus of Chapter 1 is to provide a historical fra mework f or the theory of conc epts, we w ill first to get c lear on some terminolog ical issues that ar e no t us ua lly so rt e d o ut i n th e lit e ra tur e , s o th a t th e y do no t be c ome a bu rd e n la te r. Mo st importantly , it will be helpful to provide an e x planation of wha t I take most philosophers to mean whe n they talk about conc epts and how the notion of a conc ept is relate d to other n o t i o n s s u c h a s t h o s e o f t h o u g h t s , p r o p o s i t i o n s , p r o p e r t i e s , p r e d i c a t e s , a n d m e a n i n g s . In addition, it will be important to raise the issue of va g ue pre dicates/c oncepts.

PAGE 12

3 F re g e is o ne ph ilo so ph e r w ho do e s n ot u se ' c on c e pt' in t his wa y . T he wa y I wi ll 1 use the ter m is closer to his use of the ter m 'se nse.' I do not want to be committed to say ing tha t ever y state that has r epre sentational 2 c on te nt h a s p ro po sit ion a l c on te nt. I n this diss erta tion, I will adopt the convention of putting what follows ' the 3 conce pt of' and ' the conc ept of a ' in italics. I n addition, when talking about a c oncept, I wi ll p ut t he wo rd tha t e xpre sse s th e c on c e pt i n it a lic s. I take it that when most philosophers talk about c oncepts, they are talking a bout c omm on c on sti tue nts of tho ug ht c on te nts , o r p ro po sit ion s. I a m us ing the te rm ' tho ug ht' 1 in a ver y broad se nse to mean a ny mental state with re prese ntational content (e .g ., a be lie f o r a de sir e ). A t ho ug ht c on te nt, or pr op os iti on , is ind ivi du a te d b y wh a t c on c e pts 2 are involved in it and the mode of c ombination of the conc epts. The r elations among the conce pts are not themse lves constituents of the thoug ht. Two thoug hts differ f rom one another just in case they involve either ( a) distinct conc epts with the same mode of c omb ina tio n, (b ) d ist inc t c on c e pts wi th a dif fe re nt m od e of c omb ina tio n, or (c ) t he sa me conce pts with a differ ent mode of c ombination. For e x ample, if I believe tha t triang les are trilatera l and I believe tha t triang les are triang ular, ther e ar e two distinct thoughts that I have, sinc e the c oncept of trilat eral and the c oncept of triangular are distinct concepts. 3 I f I believe tha t squares a re qua drilatera l and I believe tha t triang les are triang ular, ther e are two distinct thoughts that I have, sinc e the c oncepts of squa re a nd triang le ar e differ ent conc epts, and the c oncepts of qua drilatera l and trilatera l are differ ent conc epts. I f I believe tha t J ohn loves Mary and that Mar y loves J ohn, there are two distinct tho ug hts tha t I ha ve sin c e the mod e s o f c omb ina tio n o f t he c on c e pts a re dif fe re nt. Another distinction we ne ed to pay attention to is between toke n thoughts a nd thoughtty pes and toke n conce pts and conc ept-ty pes. We ce rtainly think that when we say that both Sue and Geor g e possess the c oncept of a tri an gle the re is s ome thi ng in

PAGE 13

4 See B urg e 1979. 4 common betwee n Sue' s conce pt of a tri an gle and Ge org e' s conce pt of a tri an gle . It wo uld be e xtre me ly dif fi c ult to m a ke se ns e of c omm un ic a tio n a mon g ind ivi du a ls a nd to ma ke g e ne ra l st a te me nts a bo ut c omm on fe a tur e s o f w ha t th e y a re thi nk ing if c on c e pts we re no t sh a re a ble c on sti tue nts of tho ug ht ( tha t is , if the y did no t po sse ss t he sa me c on c e ptty pe ). I n f a c t, i f t wo ind ivi du a ls c ou ld n ot p os se ss t he sa me c on c e ptty pe , it does not seem that we would be able to communicate with one a nother a s eff ectively as we se em to (if at a ll). I rec og nize an important sense in which sha red c onstituents of thought a re distinct: Sue and Ge org e ar e distinct individuals, and although the y share a conce pt-ty pe, their c oncepttokens must be distinct. To account f or this disti nction, one might postulate distinct mental par ticulars, but for my purposes I need not do so. I am interested in c oncepts a t the level of ty pes, and I will treat conc ept ty pes as a bstract ob je c ts. At this point we c an ra ise the following question: what is the rela tion between a conce pt being involved in someone' s thought a nd the per son g rasping the conc ept? The tr a dit ion a l vi e w i s th a t a pe rs on mus t g ra sp , o r p os se ss, a c on c e pt f or it t o b e inv olv e d in his tho ug ht. Re c e ntl y do ub ts h a ve be e n r a ise d a bo ut w he the r i n p ri nc ipl e on e ha s to possess a conc ept to have a thought that involves it. Acc ording to the traditional view, 4 in order f or one to ha ve a thoug ht, one must possess the conce pt(s) that constitute the tho ug ht ( i.e , mi nim a lly , o ne mus t be a ble to a pp ly the c on c e pt( s) c or re c tly [at le a st m os t of the time]). Suppose, for example, that I think that eg g plant is a veg etable. I n order for me to be able to have the thoug ht that eg g plant is a veg etable, I must possess the conce pts of an eggplant and a v e ge tab le (i.e., I must be able to apply them corr ectly ).

PAGE 14

5 Some philosophers arg ue that being able to apply a conc ept cor rec tly require s knowing the applica tion conditions for it. I will discuss the traditional view of c oncept possession, or g rasping a sense , when I sketch F reg e' s theory of conc epts (F reg e 1882, 1891, 1892a, 1892b, 1892c, 1893, 1918). I will give a detailed c hara cter ization of what I think concept possession comes to in Chapter 3. To us e the pr e vio us e xamp le , in e xpla ini ng my tho ug ht t ha t e g g pla nt i s a veg etable, I might use the sentenc e, ' eg g plant is a veg etable' . Sy ntactica lly , sentenc es ar e just strings of wor ds arr ang ed in a pa rticular or der a ccor ding to the r ules of a la ng uag e. Semantically , sentenc es have meaning s, which we ref er to a s propositions. I n discussing conce ptual analy sis, we ar e intere sted in the latter. A se ntence must have a me aning if I am using it to ex plain something, since when I explain som ething I purport to say something tha t is meaning ful. For example, the sentenc e ' eg g plant is a veg etable' expresses the proposition that eg g plant is a veg etable. T here is an important rela tion among propositions, sentences, a nd the meaning of a se ntence , and thoug hts, or mental repr esenta tions, which are constituted by conce pts—we have thoughts, the c ontents of which ar e propositions that are constituted by conce pts, and we e x plain our thoug hts by using se ntence s that have a meaning , or express propositions, which are constituted by conce pts. A proposition, then, is jus t the meaning of a se ntence . I nstead of ta lking solely about propositions when we talk about the mea ning of a sentenc e, one ma y talk about the mea ning of a sente nce in ter ms of the applica tion of a g ener al term to the subjec t of the sentenc e (i.e ., in terms of whe ther a predic ate is true of the ob je c t th a t th e su bje c t te rm of the se nte nc e re fe rs to) . U sin g the a bo ve e xamp le ag ain, the ter m 've g etable' is a g ener al term to which the subject of the se ntence applies (i f, in f a c t, i t is a tr ue se nte nc e ), a nd ' ve g e ta ble ' e xpre sse s a c on c e pt. Whe n w e sa y tha t a

PAGE 15

6 Vag ueness is not the only sort of indeter minacy that is releva nt when talking 5 about conc eptual ana ly sis. I will not dis cuss the issue of va g ueness a t leng th, and her e I se t it a sid e . O ne c ou ld w ri te a n e nti re dis se rt a tio n o n v a g ue ne ss a nd c on c e ptu a l a na ly sis , but that is not m y g oal her e. I n the sec tions on thought experiments, I will address other i s s u e s o f i n d e t e r m i n a c y t h a t a r i s e a b o u t w h i c h I h a v e s o m e t h i n g t o s a y. predic ate, P, expresses a c oncept, we are talking a bout P as ex pressing an ele ment of a tho ug ht, or pr op os iti on . B ut, in o rd e r f or a pr e dic a te to e xpre ss a c on c e pt a t a ll, it m us t ha ve a me a nin g . Wh e n w e sa y tha t P must have a m eaning , w ha t w e me a n is tha t P mu st ha ve a c e rt a in p ro pe rt y , a nd wh e n w e sa y tha t P att rib ute s a pr op e rty to something, we are talking a bout a fe ature of an obje ct. To fully deter mine the meaning of a se ntence , we mus t de te rm ine wh e the r t he pr e dic a te is t ru e of the su bje c t of the se nte nc e (e .g ., it must be determined w hether ' is a veg etable' is true of e g g plant). Whether a predic ate or conce pt is applicable to a pa rticular obje ct is not alway s clea r, or may be indeter minate, as we will see below. The fina l issue I would like to introduce in this section is the problem of va g ue predic ates. I addre ss the issue of vag ueness e arly in the dissertation bec ause some ma y view it as a pr oblem having to do with the application conditions (or the la ck ther eof) of predic ates, a nd, there by , the possibilit y of g etting c onceptua l analy ses. I will say at the outset that I do not think that the problem of vag ue pre dicates is a pr oblem for the pr oject of c on c e ptu a l a na ly sis . T he pr ob le m of va g ue pr e dic a te s is tha t th e re a re so me expressions whose applica tion conditions are under deter mined by our rules of use; minimally , there are some ca ses wher e our r ules for using a par ticular pre dicate do not indicate to us whe ther some pr edica te applies or f ails to apply to that case . For example, 5 it s e e ms t ha t th e ru le s f or us e of the pr e dic a te ' is b a ld' do no t always tell us whether the

PAGE 16

7 See, for example, 6 L udwig and Ray 2002. predic ate a pplies to a particula r individual or not; vag ue pre dicates do not ha ve strict application conditions, unlike some other pr edica tes. Consider the ca se of Gr andpa E d: he has re latively few strands of ha ir on the c ro wn of his he a d, bu t a bo ve his e a rs a nd a ro un d th e ba c k o f h is h e a d, he ha s q uit e a bit of hair. Clea rly , we would not asse rt, " Gra ndpa Ed is bald," as we would about someone who has a bsolutely no hairs on his hea d, or " Gra ndpa Ed is not bald," as we would about someone who ha s a full hea d of hair. O ur pre dicate ' is bald' does not tell us what to say . I n f a c t, i t se e ms t ha t tw o c omp e te nt s pe a ke rs of a la ng ua g e c a n c ome to o pp os ite conclusions about Gr andpa E d—one ca n say he is bald and the other that he is not, and our rule of use for the term doe s not indicate that one is cor rec tly apply ing the term and the other is not. Some philosophers have a rg ued plausibly that vag ue pre dicates ne ither apply nor fail to apply to any one or a ny thing a t any time, but that our use of them de termines expectations of how they apply . Those expectations we have a bout how they apply are sim ply fa lse on e s. F or e xamp le , s up po se tha t Ph il s a y s to Ma tt, " Go c a ll t he ba ld g uy in 6 for the ne x t interview." Our na tural expectation is that ' There is a bald g uy waiting to be int e rv ie we d' is t ru e . H ow e ve r, thi s e xpe c ta tio n is a fa lse on e . We ha ve fa lse expectations about vag ue pre dicates be cause in prac tice we turn a blind ey e towar d the vag ueness of vag ue pre dicates in the moment we use them (i.e., w e ac t as if there are fully specifie d application conditions for the m). We take pa rt in an eve ry day ide a liza tio n o f o ur pr e dic a te s in c on ve rs a tio n, a nd we g e t th e ha ng of the us e of the m in the same w ay . That we turn a blind ey e towar d vag ueness is unproblema tic for

PAGE 17

8 conce ptual analy sis. W hen we theorize in philosophy , we of ten (for the sake of the e nte rp ri se ) i de a lize a wa y fr om v a g ue ne ss b y pr e su pp os ing tha t th e re a re so me pr e c ise border lines or other of predic ates we use. That is, we (in a sense ) cr eate a conc ept " of the mom e nt, " wh ic h c ome s f ro m so me c omp on e nt n oti on of the pr e dic a te we a re tr y ing to analy ze, and we a naly ze it. This i s analog ous to what we do in ordina ry prac tice. I f it turns out that there a re c onceiva ble borde rline ca ses for ma ny , if not all, of our pre dicates ( and it seems that ther e ar e), one may arg ue that it is difficult to see wha t we c ou ld p os sib ly g e t, i f a ny thi ng , f ro m do ing a na ly se s o f ( va g ue ) p re dic a te s. I t is espec ially a proble m if what we are sear ching for in philosophical ana ly sis are ne cessa ry and suff icient applica tion conditions, si nce if predic ates a re va g ue, they cer tainly do not have pr ecise nece ssary and suff icient applica tion conditions. However, a s I stated above , we ide a lize va g ue pr e dic a te s in or din a ry pr a c tic e , s o it se e ms a c c e pta ble to b e do ing so in doing philosophica l analy sis, ex cepting when we are conce rned w ith vag ueness itself. Th a t is to s a y , in or din a ry pr a c tic e we us e va g ue pr e dic a te s a s if the y do ha ve fu lly specifie d application conditions. For example, we use the predic ate ' is bald' to describe c e rt a in i nd ivi du a ls, e ve n th ou g h ( str ic tly sp e a kin g ) t he pr e dic a te ne ith e r a pp lie s n or fa ils to apply to any one; and other people unde rstand wha t we mea n. Now that I have ma de cle ar my starting points and sorted out some terminolog ical issues, I will say a fe w words a bout where I think I will end up in this di ssertation and ide nti fy a re a s th a t I wi ll b e int e re ste d in . A ft e r p ro vid ing a his tor ic a l f ra me wo rk of so me theories of conce pts, I will endorse a theory of conc epts that is, for the most part, a traditional one. I say "f or the most part" beca use I think that in li g ht of what Kr ipke (1980) a nd Putnam (1975) say about natura l kind concepts, we need to modify traditional a c c ou nts of c on c e pts in c e rt a in w a y s, pa rt ic ula rl y in o ur a c c ou nt o f w ha t it is t o p os se ss a

PAGE 18

9 conce pt. After doing that, I will present a f rame work for thinking about thoug ht experiments and explain how the theory of conc epts I endorse is related to the c onditions under w hich we ma y reliably use the re sults of thought experiments as the f oundation for our philosophical theorie s. I do this by apply ing my acc ount of conc ept possession and fra mework f or thinking a bout thought experiments to the zombie thought experiment. I arg ue that we are making a mistake in our pe rfor mance of the zombie thought e xpe ri me nt, the c or re c tio n o f w hic h c re a te s a pr omi sin g op tio n f or the ph y sic a lis t. Sketch of T heories of Concepts: F rom F rege to G each I n this section, I will sketch with broad stroke s four philosophers' theories of conce pts in the history of ana ly tic philosophy : Fre g e (1882, 1891, 1892a , 1892b, 1892c, 1893, 1918), the late r Witt g enstein (1958) , Ry le (1930, 1932, 1945, 1946, 1953, 1957, 1962a, 1962b), a nd Gea ch (1956) . The re ason I use F reg e as a starting point (as opposed to, sa y , Pl a to) is t ha t I a m in te re ste d p ri ma ri ly in s tud y ing the sta te of c on c e ptu a l a na ly sis in contempora ry analy tic philosophy , and ana ly tic philosophy beg ins with Freg e. Fur thermore , Fr eg e pre serve s various fe ature s of the Platonic story about conc epts, so we need not star t from the ver y beg inning. For our purpose s, I will emphasize two featur es of theor ies of conc epts that seem to be most releva nt to conceptua l analy sis: the possession of conce pts and the deter minacy of conc ept applica tion. Major developments in philosophical thinking of c o n c e p t s u s u a l l y i n c l u d e d i s c u s s i o n o f b o t h c o n c e p t p o s s e s s i o n a n d w h e t h e r i t i s a l w a ys deter minate that a c oncept a pplies or fa ils to apply . Wit tge nstein (1958), Ry le (1930, 1932, 1945, 1946, 1953, 1957, 1962a, 1962b), a nd Gea ch (1956) can e ach be seen a s advanc ing a n alterna tive theory of conc ept possession and dete rminacy in the place of the Fr eg ean the ory . The issue of c oncept possession is ce ntral to the re st of the dissertation,

PAGE 19

10 since it play s a ce ntral role in my fra mework f or thinking a bout thought experiments. We will see that in traditional theories of conce pts it i s presumed tha t one nee d not know any thing a bout the actua l world to possess a conc ept. I n fac t, it seems as if the wa y things a re in the a ctual wor ld is entirely irrele vant to conce pt possession. I will arg ue that sometimes the way the world is is releva nt to whether w e possess a c oncept or not. T h e i s s u e o f a b s t ra ct n es s (t h at i s , wh et h er co n ce p t s ar e l o ca t ed i n s p ac e a n d t i m e) , t h o u gh c e n t r a l i n s o m e h i s t o r i c a l t h e o r i e s o f c o n c e p t s , h a s r e c e d e d i n i m p o r t a n c e ; c o n s e q u e n t l y, I will discuss it only briefly and whe n it is i mportant to a par ticular theor y of conc epts or re le va nt t o c on c e pt p os se ssi on . Wh e the r o ne thi nk s c on c e pts a re a bs tr a c t ob je c ts o r n ot, ho we ve r, do e s a ff e c t w ha t on e sa y s a bo ut c on c e pt p os se ssi on a nd ho w w e ha ve a c c e ss t o our conc epts, both of which a re r elated to wha t one say s about a prior i knowledg e and thought e x periments, which links to the re st of the dissertation. F rege Getting clea r on F reg e' s theory of conc epts will require us to ge t clear on two distinctions: (a) the distinction between F reg e' s notions of sense and r efe renc e and ( b) the dis tin c tio n b e tw e e n o ur us e of the te rm ' c on c e pt' a nd F re g e ' s, sin c e the y do no t qu ite m at ch u p . Un d er s t an d i n g t h e s en s e/ re fe re n ce d i s t i n ct i o n wi l l h el p u s s ee h o w F re ge (1882, 1891, 1892a, 1892b, 1892c , 1893, 1918) uses his term ' conce pt' ( here afte r Be gr iff ) a nd ho w F re g e ' s Be gr iff s are rela ted to our use of the term ' conce pt'. Fr eg e (1882, 1891, 1892a , 1892b, 1892c, 1893, 1918) dr aws a distinction between se ns e a nd re fe re nc e in o rd e r t o e xpla in t he int uit ive dif fe re nc e be tw e e n id e nti ty statements of the f orm ' a = a ' and ' a = b' . He a rg ues that there is more to the meaning of a n e xpre ssi on tha n w ha t it pic ks ou t ( tha t is , r e fe rs to) in t he wo rl d. As su min g tha t bo th

PAGE 20

11 Some expressions, such as ' unicorn' do not pick out any thing in the wor ld, and, 7 thus, lack a r efe rent. Such e x pressions, howeve r, do have a sense . That sense deter mines ref ere nce is a fea ture of the traditional theory of mea ning 8 that I will discuss later in connec tion with P utnam' s Twin Earth thoug ht experiment. Putnam refe rs to the re lation as the intension of an e x pression dete rmining its extension. ' a = a ' and ' a = b' are true, it seems that we can le arn something from ' a = b' that we c a nn ot l e a rn fr om ' a = a ' ; F re g e sa y s th e dif fe re nc e be tw e e n th e tw o to is a dif fe re nc e in cog nitive significa nce. F or example, consider the following two sentenc es: (1) ' the morning star = the morning star' and (2) ' the morning star = the evening star' . So long as we kn ow the me a nin g of the ide nti ty sig n, we kn ow tha t ( 1) is t ru e —th e mor nin g sta r i s identical with itself. We do not learn a ny thing ne w about the morning star. On the othe r hand, it is not im mediately clea r that (2) is true; to know that (2) is true, w e have to know that the brig htest object in the ea stern sky at dawn is, in fa ct, the same object that is the brig htest object in the we stern sky at dusk: "the differ ence betwee n the sig ns ['the morning star' and ' the eve ning sta r' ] corresponds to a diff ere nce in the mode of pres ent ati on of t he t hi ng des ignat ed" (F reg e 1892c, 152) . F re g e (1 88 2, 18 91 , 1 89 2a , 1 89 2b , 1 89 2c , 1 89 3, 19 18 ) a ttr ibu te s th e dif fe re nc e in cog nitive significa nce be tween ( 1) and ( 2) to a diff ere nce in se nse betwe en (1) and (2) . Th e se ns e of a n e xpre ssi on is i ts m od e of pr e se nta tio n, or the wa y tha t th e ob je c t th a t is the re fer ent of a n expression (if it has a re fer ent) is pre sented to us in thought. B oth 'the morning star' and ' the eve ning sta r' have the same re fer ent (that is, they both pick out 7 Venus), but they have dif fer ent senses, or meaning s. Thus, (1) a nd (2) ha ve diffe rent senses a s well. Any expression that is meaningf ul has a sense , and the r efe rent of a term is determined by its sense: 8

PAGE 21

12 G r a s p i n g a s e n s e i n F r e g e ' s t h e o r y i s a k i n t o p o s s e s s i n g a c o n c e p t i n t h e w a y I 9 ta lk a bo ut c on c e pts . The re g ular c onnection betwe en a sig n, its sense and its [refer ent] is of such a kind that to the sign ther e cor responds a definite sense and to that in turn a definite [refe rent], while to a g iven [refe rent] (an objec t) there does not belong only a sing le sig n. (F reg e 1892c, 153) What we g rasp whe n we unde rstand an e x pression is its sense. An expression may have 9 more than one sense. F or example, consider the sentenc e ' I am g oing to the ba nk' , which has at lea st two senses, one of which is that I am g oing to a financ ial instit ution of a cer tain kind, and the other of which is that I am g oing to a stretch of land at the e dg e of a body of wa ter. F or F reg e, sense s are objective in that they can be share d by more than one individual; ideas, on the other hand, ar e subjec tive: Th e sa me se ns e is n ot a lw a y s c on ne c te d, e ve n in the sa me ma n, wi th t he sa me ide a . T he ide a is s ub je c tiv e : on e ma n' s id e a is n ot t ha t of a no the r. Th e re re su lt, as a ma tter of c ourse, a varie ty of diffe renc es in the idea s associate d with the same sense . (F reg e 1892c, 154) There seems to be a ty pe/token ambig uity or conf usion here. I t looks as if ideas ar e subjective in the sense that ever y one has a differ ent token idea but they do not nece ssarily differ qualitatively . I t seems that Fr eg e would ag ree that two individuals can have the same ty pe of ide a. Tha t is, even thoug h we ha ve diffe rent e x perie nces, the re s ee m s t o b e s o m et h i n g i n co m m o n t o wh at we s ee . C o n s i d er , fo r e x am p l e, Fr ege 's telescope metaphor: Somebody observe s the Moon throug h a telesc ope. . . . The optical imag e in the telescope is indeed onesided and de pendent upon the sta ndpoint of observa tion; but it is still objective, inasmuc h as it can be used by sever al observe rs. At any ra te it c ou ld b e a rr a ng e d f or se ve ra l to us e it s imu lta ne ou sly . B ut e a c h o ne wo uld have his own r etinal imag e. (F reg e 1892c, 155) I n g e ne ra l, F re g e sa y s th a t " tho ug hts a re se ns e s o f s e nte nc e s, " a nd " [t]he tho ug ht, in itself imperce ptible by the senses, g ets clothed in the pe rce ptible ga rb of a sentenc e,

PAGE 22

13 I n " Th ou g ht, " F re g e e xpla ins c le a rl y wh a t he me a ns by ou r i nn e r w or ld o f i de a s: 10 "a w o rl d o f s en s e i m p re s s i o n s , o f c re at i o n s o f h i s i m agi n at i o n , o f s en s at i o n s , o f f ee l i n gs and mo od s, a wo rl d o f i ncl in ati on s, wi sh es an d d eci si on s" (F reg e 1918, 334). and ther eby we a re e nabled to g rasp it. We say a sente nce expre sses a th ou ght" (F re ge 1918, 328). I f a thoug ht is neither an objec t (something loc ated in spac e and time) nor an idea that is in our hea ds (that is, part of our "inne r worlds" ), then it must be an abstra ct 10 ob je c t ( tha t is , a n o bje c t no t lo c a te d in sp a c e or tim e ): A t hir d r e a lm m us t be re c og nize d. An y thi ng be lon g ing to t his re a lm h a s it in c omm on wi th i de a s th a t it c a nn ot b e pe rc e ive d b y the se ns e s, bu t ha s it in c omm on wi th t hin g s th a t it do e s n ot n e e d a n o wn e r s o a s to be lon g to t he c on te nts of his c on sc iou sn e ss (F reg e 1918, 337). So, wh e n w e un de rs ta nd the me a nin g of a n e xpre ssi on , w e g ra sp a tho ug ht. On F re g e ' s p ic tur e , th e se ns e of a se nte nc e , o r a tho ug ht, c a n b e br ok e n d ow n in to two parts—the mode( s) of pre sentation of the subjec t (or ar g ument) and the sense of the pr e dic a te . T ha t is , a se nte nc e a s a wh ole ha s a se ns e a s d o e a c h o f i ts s ub sta nti ve pa rt s (i .e ., the su bje c t a nd the pr e dic a te ). I t is wh a t F re g e thi nk s o f a s a se ns e of a pr e dic a te that I call a ' conce pt'. We c an see more c lear ly what F reg e mea ns by a sense in Fig ure 1. F ig ur e 1. F re g e ' s n oti on of a se ns e

PAGE 23

14 Th e Tr ue a nd Th e F a lse a re the on ly po ssi ble tr uth -v a lue s. 11 A s e ns e of a pr e dic a te is a c omm on c on sti tue nt o f a tho ug ht i n th e se ns e tha t th a t se ns e may be the c onstituent of many distinct Freg ean thoug hts and since these are what ar e the ob je c ts o f p sy c ho log ic a l a tti tud e s, a lso of tho ug ht c on te nts . A se ns e of a pr e dic a te is a lso an abstra ct object. We express the mode of prese ntation of a subjec t and the sense of a predic ate by a sente nce. T he subjec t term of a se ntence ref ers to an obje ct in the world, and the pr edica te term of a sentenc e re fer s to a function fr om an object to a tr uth-value. Th e re fe re nc e of a wh ole se nte nc e is a tr uth -v a lue : A statement c ontains (or a t least purports to contain) a thoug ht as its sense; and this thought is in ge nera l true or f alse; i.e., it has in g ener al a truth-va lue, which must be reg arde d as the [refe rent] of the sentenc e, just as, say , the number 4 is the [ re fe re n ce ] o f t h e e x p re s s i o n '2 +2 . ' (F reg e 1891, 139) Th e tr uth -v a lue of a se nte nc e is t he on ly fe a tur e tha t do e s n ot c ha ng e wh e n w e su bs tit ute on e te rm fo r a no the r t ha t ha s th e sa me re fe re nt. So, a ll t ru e se nte nc e s h a ve the sa me re fe re nt, a nd a ll f a lse se nte nc e s h a ve the sa me re fe re nt. 11 I n " On Con c e pt a nd Ob je c t," F re g e de fi ne s ' B e g r if f' as " the [refe renc e] of a gramm ati cal pred ica te" (F re ge 1 8 9 2 b , 1 8 2 n . 4 ). Fo r e x am p l e, co n s i d er t h e s en t en ce , 'x is a ca t'. T he pre dicate of the sentenc e is ' is a ca t', a nd the Be gr iff is the ref ere nce of the open sente nce ' x is a ca t', w hich is a func tion from an objec t to a truth-value: " if we complete the na me of a [ Be gr iff ] wit h a pr op e r n a me , w e ob ta in a se nte nc e wh os e se ns e is a th ou ght; and th is sen ten ce has a tr ut h-v alu e as i ts refere nt " (F reg e 1892a, 174) . On Fr eg e' s view, pre dicates r efe r to unsatura ted objects that a re sa turated by things to which subject terms re fer . Be gr iff s , on the other ha nd, do not refe r; they are functions from objects to truth-value s: "a function by itself must be called inc omplete, in need of su pp lem ent ati on , o r un sat urat ed" ( F r e g e 1 8 9 1 , 1 3 3 ) . F o r e xa mp le , s u p p o s e w e r e p la c e ' x'

PAGE 24

15 wi th ' Ha rr ie t' in t he a bo ve se nte nc e . ' Ha rr ie t' re fe rs to a n o bje c t, n a me ly my bla c k ta ill e ss cat, a nd ' is a ca t' r efe rs to a func tion that takes us from the objec t that is picked out by ' Har riet' and to a truth-va lue, which in this case is, as Fr eg e ca lls it , The Tr ue: " To acknow ledg e this ref ere nt as that of the Tr ue (a s the True ) is to judge that the object whi ch i s t aken as t he ar gumen t f all s u nd er t he co ncep t" (F reg e 1892a, 174) . Now that we see the rela tion between c oncepts a nd senses (na mely , that a conc ept is a sense of a pre dicate) we c an see that near ly whateve r we can sa y about one, we can say about the other . I ha ve e xpla ine d th e re la tio n b e tw e e n o ur c on c e pts a nd F re g e ' s Be gr iff s by way of explaining the sense -re fer ence distinction, and I have e x plained how F reg e ac counts for the a bs tr a c tne ss o f s e ns e s ( of pr e dic a te s) , w hic h, fo r u s, a re c on c e pts . N ow I wi ll e xpla in how Fr eg e empha siz es that it is determinate w hether a Be gr iff mov e s u s f ro m a n o bje c t to Th e Tr ue or Th e F a lse . T he re a re tw o p os sib le tr uth -v a lue s f or me a nin g fu l se nte nc e s, Th e Tr ue a nd Th e F a lse . F or a c on c e pt t ha t is e xpre sse d b y a pr e dic a te , it is d e te rm ina te wh e the r s ome thi ng fa lls un de r i t. F re g e sta te s, A def inition of a [ Be gr iff ] (o f a po ssi ble pr e dic a te ) m us t be c omp le te ; it mus t unambig uously deter mine, as re g ards a ny object, whe ther or not it fa lls under the [ Be gr iff ] (w he the r o r n ot t he pr e dic a te is t ru ly a sc ri ba ble to i t) . T hu s th e re mus t no t be a ny ob je c t a s r e g a rd s w hic h th e de fi nit ion le a ve s in do ub t w he the r i t f a lls under the [ Be gr iff ]; though f or us human being s, with our defe ctive knowledg e, the qu e sti on ma y no t a lw a y s b e de c ida ble . We ma y e xpre ss t his me ta ph or ic a lly as follows: the [ Be gr iff ] m ust have a sharp bounda ry . (F reg e 1893, 259) The re quirement that it is determinate , for a ny predic ate, P, whethe r the objec t a subject t er m re fe rs t o fa l l s u n d er P i s co n n ec t ed wi t h t h e p o s s i b i l i t y o f a n ex p re s s i o n h av i n g a sense. I stated above that for F reg e, the r efe rent of a sentenc e is a truth-va lue. Consider the open se ntence , ' x is a ca t', w hich see ms to have a se nse. I t is determinate whe ther a n ob je c t f a lls un de r t he pr e dic a te ' is a c a t' or no t. I f w ha t th e pr e dic a te ' is a c a t' a pp lie s to is

PAGE 25

16 ind e te rm ina te , th e n th e se nte nc e ' Ha rr ie t is a c a t' , d oe s n ot h a ve a re fe re nt b e c a us e in order for a n expression to have a r efe rent it must be either true or fa lse. Since moving from Ha rriet a nd the func tion 'x is a cat' to a truth-value seems to g et The Tr ue (since , of c ou rs e , H a rr ie t is a ca t), we ha ve re ason to think that it i s determinate whether something falls under the conc ept of a cat (o r t he pr e dic a te ' is a c a t' ) o r d oe s n ot. F re g e ' s r e sp on se to va g ue ne ss w or ri e s is e xpre sse d b y the fo llo wi ng : The only barr ier to enume rability is to be found in the imperfe ction of conc epts. Ba ld people for example cannot be e numera ted as long as the c oncept of ba ldn e ss is not defined so pre cisely that for a ny individual there c an be no doubt whe ther he fa lls under it. (F reg e 1882, 80) This seems to indicate that F reg e thinks that vag ue pre dicates c annot be a part of a me a nin g fu l e xpre ssi on (i .e ., on e e xpre ssi ng a se ns e ). F or F re g e , d e te rm ina te ne ss a lso pla y s a c e ntr a l r ole in h is m a the ma tic a l a nd log ic a l th e or y : . . .without complete and f inal definitions, we have no firm g round under foot, we a re no t su re a bo ut t he va lid ity of ou r t he or e ms, a nd we c a nn ot c on fi de ntl y a pp ly the laws of log ic, which c erta inly presuppose that conce pts, and rela tions, too, have sha rp boundar ies. (F reg e 1893, 265) Al tho ug h F re g e ' s Be gr iff s do not capture what I mean by ' conce pt', his sense s of predic ates do, a nd Fr eg e' s senses of pr edica tes are rela ted in an important wa y to what he c a lls ' Be gr iff s ' . "A [ Be gr iff ] is a fu nc tio n w ho se va lue is a lw a y s a tr uth -v a lue " a nd is denoted by an open se ntence (F reg e 1891, 139). The se open se ntence s may include terms denoting secondorder conce pts (that is, quantifier e x pressions), but we will not ha ve to b e c on c e rn e d w ith thi s a sp e c t of F re g e ' s th e or y . Si nc e we of te n u se se nte nc e s to e xpre ss o ur tho ug hts , w e c a n s e e tha t th e re is a re la tio n b e tw e e n F re g e ' s Be gr iff s and our conce pts (see F igur e 1). Conce pts are a t the level of thoug hts ex presse d by sentenc es and Be gr iff s a re a t th e le ve l of ob je c ts r e fe rr e d to by te rm s in se nte nc e s, wh ic h w e us e to e xpre ss t ho ug hts . F re g e ' s Be gr iff s a r e a l s o k n o w n a s ' i n t e n s i o n ' b y m a n y.

PAGE 26

17 I p u t 't h eo ry o f c o n ce p t s ' i n s ca re q u o t es as i t i s we l l -k n o wn t h at W i t t gen s t ei n 's 12 stance in the Philosophical Investigations is a nti -t he or e tic a l. I ns te a d, he pr ov ide s u s w ith a set of r emar ks about words a nd conce pts that constitute a picture of what he thinks they are and wha t role they play in lang uag e. Howe ver, f or my purposes, I will call what he h a s t o s a y a b o u t w o r d s a n d c o n c e p t s a t h e o r y. I n th is s e c tio n, I ha ve e xpla ine d h ow my un de rs ta nd ing of ' c on c e pt' is r e la te d to Fr eg e' s, what conc ept possession (in my sense of ' conce pt') comes to on F reg e' s view and Fr eg e' s commitment to t he abstra ctness and de terminatene ss of conc epts: conce pts are p a r t o f a t h i r d r e a l m o f o b j e c t s , w h i c h a r e n o t l o c a t e d i n s p a c e a n d t i m e , a n d i t i s a l w a ys d e t e r m i n a t e w h e t h e r c o n c e p t t h a t i s e x p r e s s e d b y a p r e d i c a t e a p p l i e s o r f a i l s t o a p p l y . In the next section, I will ex plain what Wittgenstein (1958) has to say about conc epts and conce pt possession. Wit tg e nst e in I n th is s e c tio n, I wi ll e xpla in Wi ttg e ns te in' s ( 19 58 ) " the or y of c on c e pts " a s it 12 appea rs in the Philosophical Investigations . A lth ou g h Wit tg e ns te in ( 19 21 ) h a s a lot to say about conc epts in the Tractatus when e x plaining his Picture T heory of mea ning, he reje cts the Picture The ory in the first third of the Philosophical Investigations . Wit tge nstein' s theory of conc epts play s a ce ntral role in the history of the study of conce pts, since it marks the beg inning of a tra dition that rejects much of what F reg e (a nd oth e rs ) e mph a size s. As I explained above, I mean by ' conce pt' w hat Fr eg e (1882, 1891, 1892a , 1892b, 1892c, 1893, 1918) me ans by ' sense of a pre dicate' , which, as I indicated a bove, we can trea t as an a bstract objec t. Fre g e introduce s senses to ac count for w hat see ms to be missing in theories of meaning wher e the mea ning of a ter m is ex hausted by its refe rent. We might say , then, that the sense of an e x pression is its meaning , since it supplies what

PAGE 27

18 Wittg e ns te in d oe s n ot s a y wh ic h w or ds ' me a nin g s a re no t e xha us te d b y the ir 13 use, so I will treat all words a s words whose me aning s are exhausted by their use. i s m i s s i n g an d d et er m i n es t h e r ef er en t (i f a n y ). Fo r W i t t gen s t ei n , h o we v er , "Fo r a l ar ge class of c lasses—thoug h not for all —in which we e mploy the word " meaning " it ca n be 13 defi ned th us : t he m eani ng of a w ord is it s u se i n a l anguage" (Witt g enstein 1958, §43). Senses of sente nces a re a lso ex plained by appea ling to use: " doesn' t the fac t that the sentenc es have the same se nse consist in their having the same us e ? " (Witt g enstein 1958, §43). The use of a w ord in a lang uag e is g overne d by rules that we lear n, and a se t of ru le s o f h ow to u se wo rd s c on sti tut e s a la ng ua g e -g a me : I n th e pr a c tic e of the us e of la ng ua g e on e pa rt y c a lls ou t th e wo rd s, the oth e r a c ts on them. I n instruction in the lang uag e the f ollowing pr ocess will occ ur: the lear ner na me s the ob je c ts; tha t is , h e utt e rs the wo rd wh e n th e te a c he r p oin ts t o the stone. . . .We ca n also think of the whole pr ocess of using w ords . . . one of tho se g a me s b y me a ns of wh ic h c hil dr e n le a rn the ir na tiv e la ng ua g e . I wi ll c a ll th ese gam es "l anguage-game s. " (Witt g enstein 1958, §7) Speaking a lang uag e just amounts to play ing a lang uag e-g ame. The re a re ma ny differ ent lang uag e-g ames, whic h corr espond to diffe rent sorts of la ng uag e use: " new ty pes of la ng ua g e , n e w l a ng ua g e g a me s, a s w e ma y sa y , c ome int o e xiste nc e , a nd oth e rs be c ome ob so let e and for gott en" (Wi ttg e ns te in 1 95 8, §2 3) . Wo rd s a re ind ivi du a te d b y the ro le they play in lang uag e, and la ng uag e is a socia l activity : "the speaking of a la ng ua g e is part of an act iv it y , o r a fo rm of l if e" (Witt g enstein 1958, §23). All of this talk about the role of w ords in a lang uag e amounts to the cla im that po sse ssi ng a c on c e pt i s n oth ing mor e tha n h a vin g a n a bil ity to u se a wo rd c or re c tly in acc ordanc e with ce rtain rules. Conce pts, then, are individuated by what role the corr esponding predic ates play in a lang uag e. F or example, consider the conc ept of a cat . The a bility to corre ctly apply the cor responding predic ate, ' is a ca t', to a particula r fur ry

PAGE 28

19 I do not think that W ittgenste in would commit him self to this, but if we ar e 14 forc ed to contra st W ittgenste in and Fr eg e, we can sa y something like this to g ive a g ener al idea of what is g oing on. cre ature in the world by following a set of r ules for a pply ing the term and the n uttering the sentenc e (pe rhaps while pointing at the fur ry cre ature ), " That is a ca t," is wha t possessing the conce pt of a cat amounts to. So, t he extent to which we use words c or re c tly is t he e xten t to wh ic h w e g ra sp c on c e pts tha t c or re sp on d to pr e dic a te s. So, it seems that for Wit tge nstein (1958), c oncept talk is under stood in terms of words, and, 14 thus, are not take n seriously as abstra ct objects. F or F reg e (1882, 1891, 1892a , 1892b, 18 92 c , 1 89 3, 19 18 ), on the oth e r h a nd , g ra sp ing a tho ug ht ( wh ic h h a s a ' c on c e pt' , in my sense of the term, a s a par t) is understanding the meaning of an e x pression that expresses the thoug ht; thoughts are abstrac t objects. I have e x plained brie fly what it is to possess a conce pt associated w ith a word on Wit tge nstein' s view and how lang uag e-g ames play a role in exhi biting conc ept possession — how well y ou play the g ame de pends on how we ll y ou know the rule s. Wit tge nstein (1958) c laims that all lang uag e use involves lang uag e-g ames a nd differ ent lang uag e use involves diff ere nt lang uag e-g ames, but that there is no single thing that we c a ll ' la ng ua g e ' (t ha t is , th e re is n o e sse nc e of a la ng ua g e ): I nstead of pr oducing something c ommon to all that we call lang uag e, I am say ing tha t th e se ph e no me na ha ve no on e thi ng in c omm on wh ic h ma ke s u s u se the sa me word for all,—but that they are related to one another in many differ ent way s. An d it is b e c a us e of thi s r e la tio ns hip , o r t he se re la tio ns hip s, tha t w e c a ll t he m a ll "lan guage." (Witt g enstein 1958, §65) As I stated above , for Wittgenste in lang uag e is a socia l activity , and whe n we spe ak a la ng ua g e we a re pla y ing a g a me . T he re is n o o ne pa rt ic ula r t hin g tha t un ite s a ll t ho se activities we c all lang uag e (i.e ., there are no nece ssary and suff icient conditions for be ing a lang uag e). Rathe r, ea ch lang uag e is similar to other things w e ca ll langua g e in var ious

PAGE 29

20 Although Wittge nstein claims that conce pts do not have nec essar y and suff icient 15 application conditions, there is a sense in whic h they do: something fa lls under a c oncept iff that thing is releva ntly simil ar to a prototy pe for the thing tha t the conce pt applies to. Of c ourse this does not under cut Witt g enstein' s point because the rig ht side of the bic on dit ion a l do e s n ot g ive us wh ic h f e a tur e s a re the re le va nt o ne s ( i.e ., ' re le va ntl y simil ar' is vag ue). 12 wa y s. Th a t is , s ome thi ng we c a ll a la ng ua g e , L , ma y be sim ila r i n r e sp e c ts R a nd R to 13 something e lse we c all a lang uag e, L ' , but L may be similar in respe cts R and R to even 23 so me thi ng e lse we c a ll l a ng ua g e , L ' ' , a nd in r e sp e c ts R a nd R to L ' ' ' : W e s ee a c o m p l i ca t ed n et wo rk o f s i m i l ar i t i es o v er l ap p i n g an d cr i s s -c ro s s i n g: sometimes overa ll sim ilarities, sometimes simi larities of de tail. I can think of no better e x pression to cha rac terize these similarities than " family rese mblances" ; for the var ious resemblanc es betwe en member s of a f amily : build, feature s, color of ey es, g ait tempera ment, etc. e tc. over lap and c riss-cr oss in the same wa y .—And I shall say : 'g ames' form a f amily . (Witt g enstein 1958, §66-7) F o r W i t t g e n s t e i n , i t l o o k s a s i f m o s t , i f n o t a l l , o f o u r c o n c e p t s a r e w h a t I w i l l c a l l f a m i l yr e s e mb la n c e c o n c e p ts . T h e s e a r e c o n c e p ts th e a p p li c a ti o n o f w h ic h w e c a n c h a r a c te r iz e in s ome thi ng a lon g the fo llo wi ng lin e s: x f a lls un de r t he c on c e pt C if f x be a rs a c e rt a in fa mil y re se mbl a nc e to o the r t hin g s th a t f a ll u nd e r t he c on c e pt C . No te tha t th is i s n ot t o 15 g iv e a p p li c a ti o n c o n d it io n s ; i t i s to c h a r a c te r iz e h o w w e u s e a n d d e p lo y c o n c e p ts . We c a nn ot g ive a pp lic a tio n c on dit ion s b e c a us e it u se s ' c on c e pt C ' on the rig ht side and invokes the notion of fa mily rese mblance w ithout specify ing w hat those re semblance s need to be . C , then, (a s we will see be low) is understood in terms of a prototy pe; whethe r something f alls under C de pe nd s o n h ow sim ila r t ha t th ing is ( in t he re le va nt r e sp e c ts) to the prototy pe in terms of w hich C is understood and other things that fa ll under the conce pt of C . For example, consider the c oncept of a bir d , w ith wh ic h w e of te n a sso c ia te wi th t he fo llo wi ng c ha ra c te ri sti c s: h a s f e a the rs , c a n f ly , la y s e g g s, c hir ps , e tc . O f c ou rs e not ever y bird meets all of the se cr iteria, but ce rtainly ever y bird must meet some (or most) of them.

PAGE 30

21 We should note here tha t although r ules g overn the use of wor ds, there mig ht 16 not be a r ule for e very situation in which we ar e conside ring using a word. I put 'de finitions' in sca re quote s here beca use Witt g enstein does not use the 17 t e r m a s w e u s u a l l y d o i n p h i l o s o p h y. So, the way that we tell whe ther something is a g ame is by looking to a pr ototy pe of a g a me a nd oth e r g a me s a nd the n d e te rm ini ng to w ha t e xten t th e thi ng in q ue sti on is sim ila r t o th a t pr oto ty pe a nd oth e r g a me s: Ho w s ho uld we e xpla in t o s ome on e wh a t a g a me is? I ima g ine tha t w e sh ou ld descr ibe games to h im, a nd we mig ht a dd : " Th is and similar thi ngs are called 'g ame s. '" (Witt g enstein 1958, §69) I take the ' this' he re to be ref err ing to a prototy pe, or a n example, of a g ame. A pr ototy pe is supposed to give us a roug h idea of when we should and should not apply a conc ept (use a word); it g ives us g ener al rules on how to use terms, but that is not to say that 16 wo rd s d o n ot h a ve me a nin g s: " a re we to s a y tha t w e do no t r e a lly a tta c h a ny me a nin g to t h i s w o r d , b e c a u s e w e a r e n o t e q u i p p e d w i t h r u l e s f o r e v e r y p o s s i b l e a p p l i c a t i o n o f i t ?" (W i t t gen s t ei n 1 9 5 8 , § 8 0 ). Fo r W i t t gen s t ei n , wo rd s h av e m o re o r l es s ro u gh "de finitions." 17 I stated above that for Wittgenste in (1958) conc ept possession amounts to being a ble to u se a wo rd c or re c tly (a nd in a c c or da nc e wi th c e rt a in r ule s) , b ut I did no t st a te wh a t e xac tly Wittg e ns te in t hin ks c on c e pts are . Given that Wittgenstein thinks the meaning of a w ord is exhausted by its use, it seems that when we ask a que stion about how we a re to use se ntence s in which ' conce pt' a ppear s, we should ask how the expression is used. That is, on this vi ew, it is not appropria te to ask the que stion, "What is a c on c e pt?" be c a us e it d oe s n ot m a ke se ns e to t a lk a bo ut c on c e pts ; Witt g e ns te in o nly wi ll t a lk a bo ut t he us e of the wo rd ' c on c e pt' . Si nc e the re a re no c on c e pts (t a ke n s e ri ou sly as entities) on Wittgenstein' s view, conc epts are cer tainly not abstrac t objects.

PAGE 31

22 In the first section of this chapter, I stated that if concepts were not shareable constituents of thought (at the level of types), then it would be difficult to communicate with one another effectively. From what Wittgenstein (1958) says, it seems that since two individuals can use the same words in the same way, there is a sense in which we can share concepts, which are understood in terms of a prototype. Rules for the use of a word are partial in the sense that there is not for every situation a rule about whether the word applies or fails to apply, and we may think in terms of different prototypes. Since language is a social activity it seems likely that the prototypes that each of us appeal to bear certain family-resemblance relations to one another just as they do to those things that fall under the prototype in terms of which the concept is understood. Now that we have a general picture of Wittgenstein's "theory of concepts," we are in a position to see how Wittgenstein (1958) would respond to the (apparent) problem of vague predicates. In the previous section, we saw that for Frege (1882, 1891, 1892a, 1892b, 1892c, 1893, 1918) vagueness is not a problem at all. If a predicate that allegedly expresses a concept is vague, then it does not express a concept at all: Frege compares a concept to an area and says that an area with vague boundaries cannot be called an area at all. This presumably means that we cannot have anything to do with it.—But is it senseless to say, "Stand roughly there"? Suppose that I were standing with someone in a city square and said that. As I say it, I do not draw any kind of boundary, but perhaps point with my hand—as if I were indicating a particular spot . . . . One gives examples and intends them to be taken in a particular way.—I do not, however, mean by this that he is supposed to see in those examples something in common which I—for some reason—was unable to express. (Wittgenstein 1958, ) Vagueness is not a problem for Wittgenstein, either, because according to his "theory of concepts," the meaning of a word is to be sought in its use. And there is no reason to think that the use of a word must determine for all conditions whether it is to apply or not. Correspondingly, since for Wittgenstein talk of concepts expressed by words is to be spelled out also in terms of the use of those words (since grasping the concept expressed

PAGE 32

23 by a word is knowing its use in the language), it will turn out that concepts need not have necessary and sufficient application conditions. That is, for Wittgenstein, we "associate concepts" with all words that have a use in the language (that is, have a meaning), and then since the practices governing use can be partial, so may the application conditions for concepts. When we use words, we are playing a language-game. The rules of the language-game, which tell us when to use words, may be partial in the sense that the rules that tell us how to use words do not tell us all of the circumstances in which a particular word applies. Furthermore, sometimes there is no rule that tells us whether or not a word applies in a given situation: "the application of a word is not everywhere bounded by rules" (Wittgenstein 1958, ). Rather, the rules serve as a partial guide, and we apply a predicate just in case whatever we want to apply the predicate to bears a certain family resemblance to an appropriate prototype. In this section, I explained how Wittgenstein (1958) rejects much of what Frege (1882, 1891, 1892a, 1892b, 1892c, 1893, 1918) had to say about concepts. We learn a language by having someone point out to us things that are of certain kinds. We are shown examples, or prototypes, of the kind, and, when deciding whether to use a particular term, we judge whether the object we want to apply the term to bears certain family resemblances to the prototype. However, there is not always a fact about the matter whether a term applies in a situation or not. That is, our rules for the use of a term or the representations we use to judge whether something falls under a concept do not always determine whether the concept applies. In the next section, I will explain Ryle's theory of concepts, which is an intermediary one between Frege's and Wittgenstein's. Ryle Ryle's theory of concepts goes through several stages of development, but most of the important things he wants to say remain relatively constant. His work on concepts

PAGE 33

24 begins with a general skepticism of using the term 'concept' in any expression because, he argues, all talk about concepts is misleading and, thus, should be replaced by talk about specified general terms. In his later work, he is more tolerant of the use of the term 'concept', but, like Wittgenstein (1958), he does not think that concepts are abstract objects. In this section, I will sketch what I take to be the most general features of his theory. In "Systematically Misleading Expressions," Ryle claims that any expression containing 'concept' is a misleading one and should be replaced by talk about general terms: Sometimes philosophers say that they are analysing or clarifying the 'concepts' which are embodied in the 'judgments' of the plain man or the scientist, historian, artist or who-not. But this seems to be only a gaseous way of saying that they are trying to discover what is meant by the general terms contained in the sentences which they pronounce or write. For, as we shall see, ' x is a concept' and ' y is a judgment' are themselves systematically misleading expressions. (Ryle 1932, 39) There are two things that seem to be going on here: first, Ryle is rejecting the view that concepts, or what talk about concepts can be replaced with, are abstract objects, and, second, he is claiming that it does not make sense to talk about analyzing concepts. For these reasons, it is misleading to talk about concepts. We will see the former more clearly when we talk about what Ryle thinks meanings are. If concepts are "what is meant by the general terms" in sentences, then for someone to use and/or to understand a sentence containing those general terms "intelligently," he or she needs to know what those general terms mean. That is, there is no question about whether one understands what he or she is saying when using a sentence intelligently. If one is not using a sentence intelligently then one is "just gabbling parrot-wise" (Ryle 1932, 39).

PAGE 34

25 Although in Ry le' s later wor k he ac cepts talk of c oncepts, his view a bout what they are does not cha ng e. I n "Phenome nology Ver sus "The Concept of Mind," " he de fi ne s ' c on c e pt' , b ut p re se rv e s w ha t he sa y s in so me of his e a rl ie r w or k a bo ut c on c e pts being what is meant by a g ener al term in a se ntence : B y ' c on c e pt' we re fe r t o th a t w hic h is sig nif ie d b y a wo rd or a ph ra se . I f w e ta lk of the c oncept of Euclidean point we a re re fe rr ing to w ha t is c on ve y e d b y thi s Eng lish phrase, or by any other phra se, Gre ek, F renc h, or Eng lish, that has the same mea ning. (Ry le 1962a, 182) I f all talk about c oncepts should, or a t least could, be r eplac ed by talk about what is meant by the terms in sentenc es, then in orde r to deter mine what Ry le' s theory of conce pts is, we have to e x amine wha t he say s about meaning (s). So far , it is not clea r whether Ry le' s use of the te rm ' conce pt' ma tches up with wha t I call ' conce pts' ( i.e., common constituents of thoug hts). Clearly , Ry le thinks that differe nt expressions can ha ve the sa me me a nin g . So , d if fe re nt s e nte nc e s ma y be sa id t o " e mpl oy " the sa me c on c e pts . D if fe re nt s e nte nc e s c on ta in d if fe re nt w or ds tha t e xpre ss t he sa me c on c e pts , and since thoughts a re e x presse d by sentenc es, we can sa y that differ ent thoug hts can inv olv e the sa me c on c e pts . I n " Th ink ing Th ou g hts a nd Ha vin g Con c e pts ," he sta te s: Th e me a nin g s o f t he wo rd s in a se nte nc e a re fu nc tio na lly int e rl oc kin g . Co nc e pts are not bricks out which truths and f alsehoods ar e built. . . . Rather a c oncept, sa y the conc ept of if or uncle of , is what ca n be common to an indef inite rang e of oth e rw ise dif fe re nt i nte g ra l tr uth s, fa lse ho od s, qu e sti on s, c omm a nd s, e ntr e a tie s, etc. (Ry le 1962b, 449) I ha ve a lr e a dy sta te d th a t Ry le do e s n ot p re se rv e the fe a tur e of a bs tr a c tne ss i n h is theory of conc epts. I should point out here that there seem to be two thing s we c an mea n wh e n w e ta lk a bo ut c on c e pts be ing a bs tr a c t ob je c ts, on e of wh ic h is so me thi ng we sh ou ld not want to be committed to and the other of which w e nee d not worry about. Fir st, there is t he F re g e a n v ie w a c c or din g to w hic h c on c e pts a re no t lo c a te d in sp a c e a nd tim e , b ut i n

PAGE 35

26 a thi rd re a lm. Se c on d, the re is t he vie w a c c or din g to w hic h c on c e pts a re ty pe s o f t hin g s, which is something w e nee d not worry about being committed to. 'Ty pe' in the sense I a m ta lki ng a bo ut h e re is s ome thi ng tha t a no min a lis t c a n a c c e pt, a nd it i s c omp a tib le wi th a prototy pe. Ry le explicitl y reje cts conc epts as abstra ct objects a nd philosophy as the st ud y of m eani ngs (c on si dere d as abs tr act ob jec ts ) i n "Th e Th eory of M eani ng": I f the mea ning of an expression is not an entity denoted by it, but a sty le of opera tion perfor med with it, not a nominee but a role , then it is not onl y repe llent bu t po sit ive ly mis le a din g to s pe a k a s if the re e xiste d a Th ir d Re a lm w ho se denizens are Meaning s. We can distinguish this knight, a s a piec e of ivory , from the par t it or any proxy for it may play in a g ame of chess; but the par t it may play is not an ext ra e ntity , made of some my sterious non-ivory . . . . To say , there fore , tha t ph ilo so ph y is t he sc ie nc e of Me a nin g s, tho ug h n ot a lto g e the r w ro ng , is lia ble to misl ead. (Ry le 1957, 370-371) We have g otten a little bit more here about what Ry le thinks about meaning s than just that they are not abstrac t objects as F reg e thinks they are . Ry le ar g ues that to fig ure out the meaning of an e x pression one ha s to figur e out the role the expression play s in a lang uag e. The role a n expression play s in a lang uag e is ana logous to the r ole a knig ht play s in a che ss ga me. Obviously the role of the ivory object is not something a longside the ivory object. So, the role of an expression in a lang uag e is not something a longside the words that make up the expression. This sounds much like what Wit tge nstein (1958) thinks meaning is—the meaning of an e x pression is how it is used in a lang uag e. For Wit tge nstein speaking a lang uag e is a g ame whe re the rules ar e par tial in a cer tain way . I n some way s, Ry le adopts this kind of view as we ll, but he seems to think of spea king a lang uag e as pla y ing a g ame in a loose r, metaphor ical, sense . I n "O rdinary L a ng ua g e ," he sta te s, I f I know the mea ning of a wor d or a phr ase I know something like a body of unwritten rules, or something like a n unwritten code or g ener al re cipe. I have lear ned to use the w ord cor rec tly in an unlimited variety of diffe rent setting s. Wha t I kn ow is, in t his re sp e c t, s ome wh a t li ke wh a t I kn ow wh e n I kn ow ho w t o

PAGE 36

27 use a knig ht or a pa wn at che ss. I have le arne d to put it t o its work any when a nd any wher e if ther e is work f or it to do. (Ry le 1953, 312) Given wha t Ry le say s here , to understand the me aning of a w ord or a phrase we have to know c erta in rules that g overn the use of the w ord. Unlike Wittgenstein, howeve r, it seems her e that whe ther a word a pplies to a situation or not is determinate. The rule s for using knights a nd pawns in a c hess g ame a re stric t—in any g iven situation in a che ss ga me, there is a fac t of the matter a bout what a knig ht or a pa wn ca n do. Ry le' s commitment to determinateness is also sug g ested by what he sa y s in "The Theory of Mea ning," wher e he dr aws a connec tion between know ing the meaning of an expression and knowing how to use an e x pression in a lang uag e: To know wha t an expression means involves knowing what ca n (log ically ) be said with it and what ca nnot (log ically ) be sa id from it. I t involves knowing a set of bans, f iats and oblig ations, or, in a wor d, it is t o know the rule s of the employ ment of that expression. (Ry le 1957, 363) So, on e of the wa y s in wh ic h Ry le dif fe rs fr om Wi ttg e ns te in i s th a t Ry le se e ms n ot t o allow for the kind of partiality in the rules for using e x pressions that Witt g enstein does. For Ry le, there is more to g rasping the meaning of a w ord than just knowing the sufficie nt conditions for its application; gr asping the meaning of a w ord re quires kn ow ing its ne c e ssa ry a nd su ff ic ie nt a pp lic a tio n c on dit ion s: A boy who has lea rned to c ount and is now beg inning a rithmetic is told that numbers divide up into odd numbers and e ven number s. He a sks which ar e which and is told, fir st , that 2, 4, 6, 8 and 10 ar e eve n numbers, while 1, 3, 5, 7 or 9, a re od d n umb e rs ; second , th a t th e nu mbe r o f t he le ft -h a nd pa g e of a ny bo ok is an eve n number. . . . Af ter his fourth lesson he can te ll us not onl y what number s a re e ve n b ut a lso , s o to sp e a k, wh a t make s them even. A nd, now we a sk, at which stag e has he ' g ot the conce pt of eve n number ' ? F or a t e a c h s ta g e the re is s ome t as k wh i ch h e c an p er fo rm , wh i ch h e c o u l d n o t d o b ef o re , an d b y t h e t h i rd s t age he ca n corr ectly answe r all questions of the f orm ' I s . . . an eve n number? ' Yet we are inclined to say that so long a s he has not y et re alised that eve n numbers ar e divisibl e by 2, he has not re ally g ot the conce pt of eve n number ; or else that he has not y et g ot the whole of it. (Ry le 1962b, 450)

PAGE 37

28 I t looks as though on this view knowing nece ssary and suff icient applica tion 18 conditions is necessar y , but not sufficient for g rasping a conc ept. That is, knowing what ma ke s a c on c e pt a pp ly se e ms a bo ve a nd be y on d k no wi ng its a pp lic a tio n c on dit ion s. I am re placing ' proposition' with ' sentenc e' here , beca use in " Are There 19 Propositions ? " ( 1930), Ry le states, " The na me ' proposition' then de notes, what g rammar ians have alway s used it to denote, the same as ' sentenc e' or ' statement.' . . .' Proposition,' then, is a na me not for wha t I think but for what I think and talk in " (R y le 1930, 37). He re R y l e's co m m i t m en t t o d et er m i n at en es s b ec o m es v er y cl ea r: I f gr as p i n g a c on c e pt r e qu ir e s k no wi ng ne c e ssa ry a nd su ff ic ie nt a pp lic a tio n c on dit ion s, the n Ry le mus t a lso thi nk tha t c on c e pts ha ve ne c e ssa ry a nd su ff ic ie nt a pp lic a tio n c on dit ion s. I f c on c e pts have ne cessa ry and suff icient applica tion conditions, t hen ther e is a fa ct of the ma tter whether a conc ept applies or f ails to apply in a par ticular situation. Henc e, like F reg e, he pr e se rv e s th e de te rm ina te ne ss o f c on c e pts . I t is no t e no ug h to be a ble to u se a wo rd in a ll o f t h e r i g h t s i t u a t i o n s ( i . e . , o n e h a s t o k n o w n o t o n l y that a wo rd a pp lie s, bu t w ha t make s it apply ). 18 Al t h o u gh R y l e d o es n o t s ee m t o ad d re s s t h e i s s u e o f v agu en es s , fr o m o t h er t h i n gs he sa y s, it l oo ks a s if he wo uld sa y tha t if a pr e dic a te is v a g ue , it do e s n ot e xpre ss a conce pt. Call thi s semantic vag ueness. How ever , Ry le ar g ues that our unde rstanding of when to use w ords is alway s incomplete: As pe op le ' s u nd e rs ta nd ing of the [se nte nc e s] tha t th e y us e is a lw a y s im pe rf e c t, 19 in the sense that they never have r ealized and ne ver c ould rea liz e all the log ical po we rs of tho se [se nte nc e s], s o th e ir g ra sp of ide a s o r c on c e pts is n e c e ssa ri ly incomplete. (Ry le 1945, 200) Ca ll t his e pis te mic va g ue ne ss. Ep ist e mic va g ue ne ss a pp lie s to e ve ry wo rd , w he the r i t is vag ue or not. So, there is no semantic vag ueness of predic ates be cause any predic ate that is v a g ue do e s n ot e xpre ss a c on c e pt, a nd e pis te mic va g ue ne ss i s n ot a pr ob le m be c a us e

PAGE 38

29 even if we ne ver f ully g rasp the ne cessa ry and suff icient applica tion conditions for our c on c e pts , it do e s n ot m e a n th a t th e y do no t ha ve the m. Ge ach I n this section, I will sketch Gea ch' s theory of conc epts as it appea rs in Mental Ac ts sin c e mos t, i f n ot a ll, of Ge a c h' s w or k o n c on c e pts is c on ta ine d in it (Ge ach 1956) . L ike Witt g enstein and Ry le, Gea ch empha siz es that lang uag e is ce ntral to talk about conce pts and reje cts the abstra ctness of c oncepts. Unlike Wittge nstein and Ry le, howeve r, Ge ach doe s not think t hat spea king a lang uag e is nec essar y for possessing c on c e pts . For Gea ch, " [c] oncepts, a s we shall see , are capa cities exercised in a cts of judgme nts—psy cholog ical conc epts, in psy cholog ical judg ments about oneself and ot hers " (G e a c h 1 95 6, 7) . A n a c t of jud g me nt i s p utt ing on e ' s b e lie f i nto wo rd s " wi th con si dera ti on " (Ge ach 1956, 9) . We will see below that a n act of judgme nt does not require an ability to put a belief into words. F or now, howe ver, I will talk about making jud g me nts by sp e a kin g a la ng ua g e . Wh a t it me a ns fo r a c on c e pt t o b e a c a pa c ity ex erci sed in an act of j ud gment , t hen , i s fo r us to be ab le t o ex pres s a b eli ef s uch as, "I believe tha t it is war m outside," whe n we be lieve that it is warm outside. Since "[c]oncepts . . . ar e pre supposed to, and exerc ised in, acts of judg ment," c oncepts c an be sa id t o b e c on sti tue nts of tho ug hts (Ge ach 1956, 11) . That is, the ability to have a thought pr ece des the a bility to express that thought (e .g ., in order to truly say , "I believe that it is warm outside," w e have to be able to ha ve the thoug ht that it is war m outside). Pos se ssi ng a c on c e pt i s a c a pa c ity to u se a n e xpre ssi on c or re c tly a nd tha t c a pa c ity is g ro un de d in be ing a ble to h a ve a tho ug ht. I n th is s e ns e , w e ma y sa y tha t a c on c e pt i s a constituent of a thoug ht. Geac h does not, howeve r, think that conce pts are a bstract

PAGE 39

30 Wha t G e a c h d oe s n ot m a ke c le a r i s h ow we c ome to b e a ble to h a ve tho ug hts in 20 the first plac e and how we a cquire the ca pacity to make judg ments. That is, Gea ch claims that in order to corre ctly apply a wor d, we must alre ady (in a sense ) possess the conce pt, but he does not make it cle ar how we g rasp it, and he does not tell us how we acquir e the a bility to make ce rtain judg ments. objects as F reg e does—c oncepts for Gea ch ar e primar ily abilities to use expressions corr ectly , or ca pacities that we exercise whe n we make judgme nts. W e mig ht say conce pts are a bstract in the se nse that they are a ty pe of c apac ity , but this i s an un my ste ri ou s k ind of a bs tr a c tne ss. Gea ch explains the relation betwe en the c orre ct use of w ords (a nd sentenc es) a nd c on c e pt p os se ssi on a s f oll ow s: The a bility to express a judg ment in words thus presupposes a number of capa cities, previously acquir ed, for intelligently using the sever al words a nd ph ra se s th a t ma ke up the se nte nc e . I sh a ll a pp ly the te rm " c on c e pts " to t he se sp e c ia l c a pa c iti e s— a n a pp lic a tio n w hic h I thi nk lie s f a ir ly c los e to t he his tor ic us e of the ter m. I t will be a sufficient condition for James's ha ving the conce pt of soan dso tha t he sh ou ld h a ve ma ste re d th e int e lli g e nt u se (i nc lud ing the us e in madeup sentenc es) of a wor d for so -a nd -s o in s ome la ng ua g e . T hu s: i f so me bo dy kn ow s h ow to u se the En g lis h w or d " re d" , h e ha s a c on c e pt o f r e d; i f he knows how to use the first-per son pronoun, he ha s a conc ept of se lf ; if he knows how to use the ne g ative construc tion in som e lang uag e, he ha s a conc ept of neg ation. (Ge ach 1956, 1213) To intellige ntly use words a nd phrase s that make up a sentenc e, one must first be able to have thoughts involving the conc epts that are expressed by the words in the sentenc e. Howeve r, it is not necessar y that one be a ble to master the intellig ent use of a 20 word that e x presse s a conc ept in order to possess that conce pt. That is, there a re othe r way s one mig ht possess a conc ept (i.e., the re a re othe r suffic ient conditions for conc ept possession). Th a t c on c e pt p os se ssi on do e s n ot r e qu ir e the a bil ity to s pe a k a la ng ua g e is a notable diffe renc e betwe en Ge ach' s theory of conc epts and Wittgenstein' s and Ry le' s.

PAGE 40

31 I n the ca se of the a phasia pa tient, perha ps the individual ex presse s his belief that 21 it is warm outside by fanning his face . For Gea ch the a bility to use a wor d corr ectly is sufficient, but not nece ssary for c oncept possession. Wi ttge nstein and Ry le, on the other hand, ar g ue that talk about c oncepts c an be re du c e d to ta lk a bo ut t he us e of a wo rd a nd wh a t is me a nt b y a g e ne ra l te rm , re sp e c tiv e ly . G e a c h s ta te s, I t would be har d to devise a nonverba l criteria for the [apha sia] patient's ha ving reta ined a c oncept of the day after tomorrow . The c entra l and ty pical applica tions of the ter m "ha ving a conce pt" a re those in whic h a man is master of a bit of linguistic usag e; we c an then r easona bly extend the term to case s sufficiently like these. (Ge ach 1956, 13) Gea ch ac knowledg es the diff iculty in determining what other criter ia than ver bal ones would be suf ficient c onditions for conce pt possession. However , he does not wa nt to s a y tha t th e re a re no su ff ic ie nt c ri te ri a fo r c on c e pt p os se ssi on tha t is no t g ro un de d in some ability to speak a lang uag e. So, if someone maste rs the use of a wor d, then we ma y sa y tha t he or sh e po sse sse s th e c on c e pt e xpre sse d b y tha t w or d. B ut i f s ome on e is unable to use w ords but can e x hibit som e beha vior that is simi lar, for example, ostensively exhibi ting a capa city to make a judg ment about something , we should not 21 rule out the possibility that the individual possesses ce rtain conc epts. I stated ea rlier in the c hapter that if conc epts wer e not share able c onstituents of thought then two individuals could not possess the same conce pt in the same wa y that tw o in div idu a ls c a nn ot h a ve on e a nd the sa me bo ok a t th e sa me tim e . I n o ne se ns e conce pts are not sha rea ble for G eac h since e ach individual has diff ere nt token capa cities of the same ty pe. Howe ver, he arg ues that, in fac t, more than one person c an possess the sa me c on c e pt:

PAGE 41

32 A conc ept, as I am using the term, is subjective—it is a mental c apac ity belong ing to a par ticular pe rson. . . . The subjec tive nature of conc epts does not howeve r im pl y th at i t i s i mp rop er t o s peak of t wo p eop le a s "ha vi ng th e sam e con cept s"; confor mably to my explanation of the term " conce pt", this will mean that they have the same menta l capa city , i.e. ca n do essentially the same thing s. (Ge ach 1956, 14) Concepts ar e ca pacities that we exercise in making judgme nts. W hen we speak of two thought c ontents as involving the sa me conc epts, what we mean is that the judg ment involves the same c apac ity . Two individuals that have the sa me conc ept, then, ar e sim ply us ing the sa me c a pa c ity in m a kin g so me jud g me nt. Gea ch does not think that we a cquire conce pts by observing fea tures of thing s and a bs tr a c tin g fr om t ho se fe a tur e s: W e c an n o w s ay s o m et h i n g t h at go es fo r a l l co n ce p t s wi t h o u t ex ce p t i o n : Ha v i n g a c on c e pt n e ve r m e a ns be ing a ble to r e c og nize so me fe a tur e we ha ve fo un d in direc t experience ; the mind make s conce pts, and this conceptformation and the subsequent use of the conc epts forme d is never a mere rec og nition or finding; but this does not in the least preve nt us from apply ing c oncepts in our se nsee xpe ri e nc e a nd kn ow ing so me tim e s th a t w e a pp ly the m r ig htl y . I n a ll c a se s it is a matter of f itting a c oncept to my experience , not of picking out the fea ture I am interested in f rom among other fe ature s g iven simultaneously . (Ge ach 1956, 40) L ater , Gea ch make s it clear w hat he mea ns when he say s that conce pts are not only a re cog nition or finding: " no conce pt whatsoeve r is to be identified with a reco gnit io nal capac it y " (G ea ch 1 9 5 6 , 7 7 ). Ha v i n g a r ec o gn i t i o n al ca p ac i t y i s n o t en o u gh to be able to make an intellige nt judgment. Althoug h Gea ch states that " the mind make s c on c e pts ," he do e s n ot m a ke it c le a r j us t ho w i t do e s (Ge ach 1956, 40) . I t seems that he wants to say that althoug h we may knowing ly rig htly apply sensory conce pts, we do not acquir e sensor y conce pts merely by having sensory experience s. This is in contrast to the abstrac tionist view, ac cording to which " a conc ept is acquire d by a proc ess of sing ling out in attention some one fea ture g iven in direc t experience — abstracting it—and ignor ing the other fe ature s simult aneously g iven— abstracting fro m t hem " (Ge ach 1956,

PAGE 42

33 An abstra ctionist about conce pt acquisition need not be c ommitt ed to a 22 particula r ontolog y (i.e., she ne ed not believe that conce pts are a bstract objec ts). Howeve r, it may seem na tural to explain abstractionism as follows: when we abstrac t from experienc e ce rtain fe ature s of an e x perie nce, w e g rasp a conce pt. The conc ept we g rasp is an obje ct not located in spa ce a nd time that has those fe ature s we a bstracte d from o u r e x p e r i e n c e . C o n c e p t s a r e m o r e c o a r s e g r a i n e d t h a n o u r e x p e r i e n c e s . A l t e r n a t i v e l y, we c an think of them by abstrac ting them f rom our experienc e; we g et a pr ototy pe, which is a lon g the sa me lin e a s Wit tg e ns te in t hin ks of c on c e pts . I wa nt t o p oin t ou t he re so me thi ng tha t I dis c us s b ri e fl y la te r i n Ch a pte r 5 : if 23 conce pts are not the a bility to rec og nize some featur e of dire ct experienc e, then ther e ar e no rec og nitional concepts. Tha t is, accor ding to Ge ach, ha ving a conce pt is a complex capa city . I f this is right, he w ill reject re cog nitional concepts discussed in L oar 1990. 24 Sellars (1997) sounds a lot like Geac h does her e. 18). Although it is important to ge t clear on what Ge ach me ans by ' rec og nition', he 22 do e s n ot ( se e m to ) d e fi ne it a ny wh e re . T his is a pr ob le m si nc e it i s u nc le a r w he the r i t is just bringing some fea ture of e x perie nce unde r a c oncept or if it involves subsumi ng thi ng s u nd e r c on c e pts . M ini ma lly , w e c a n s e e tha t it sta nd s in so me re la tio n to a bs tr a c tio nis m, s inc e he ta lks a bo ut c on c e pts no t be ing a me re re c og nit ion wh e re he is arg uing a g ainst abstrac tionism . So, perhaps wha t he means by ' rec og nition' is along the same lines as w hat happe ns befor e we make a n abstrac tion from experience to acquire a conce pt on the abstrac tionist view. 23, 24 Unlike Witt g enstein, but like Ry le, Gea ch is silent on whether he thinks that deter minateness is a fe ature of conc epts. Given wha t he say s, howeve r, about how c on c e pts a re c a pa c iti e s to ma ke c e rt a in j ud g me nts , it loo ks a s if c on c e pts a re de te rm ina te (i .e ., the re is a lw a y s a fa c t of the ma tte r a bo ut w he the r a c on c e pt a pp lie s) . I f c on c e pts we re no t de te rm ina te , it wo uld be dif fi c ult to s a y wh a t th e c a pa c ity to m a ke jud g me nts containing a conc ept amounts to. That is, if cer tain words do not have nece ssary and sufficie nt application conditions (that is, if they do not have strict bounda ries), the n the capa city associate d with the use of a word doe s not amount to possessing a c oncept. F or

PAGE 43

34 it s e e ms a s th ou g h th e c a pa c ity to m a ke c e rt a in j ud g me nts a bo ut t he us e of a wo rd ha s to be g rounded in something like strict application conditions (in some wa y ). Thus, vag ue predic ates do not express conc epts. I f Ge ach ha d some kind of prototy pe view of conce pts, as Wit tge nstein does, then we might be a ble to make se nse of c oncepts not having sharp bounda ries. Howe ver, it does not see m that Geac h has a pr ototy pe view of conce pts beca use he thinks that we ha ve to have the conc ept bef ore w e ca n make jud g me nts a bo ut o ur e xpe ri e nc e . A pr oto ty pe vie w o f c on c e pts se e ms t o b e g ro un de d in the view that c oncept possession involves first " seeing " a n example of something in the world that fa lls under the conc ept and then doing some kind of abstra cting of a pr ototy pe to apply it to som ething in the world. Co nt e m por ar y The or ie s o f Co nc e pt s: Be ale r and P e ac oc ke I n this section, I will sketch two prominent contempor ary theories of conce pts by Be aler (1987, 1992, 1998b, 2002a) and Peac ocke ( 1989, 1992), pay ing c are ful attention to h ow the y fi t in wi th t he the or ie s I dis c us se d in the pr e vio us se c tio n. I n p a rt ic ula r, I wi ll use the same fra mework, e x plaining how eac h theory acc ounts for conc ept possession and the de terminac y of conc ept applica tion. As with some hist orica l theories of conce pts, contemporar y developments in philosophical thinking of conce pts can be seen as trea ting c oncept possession and de terminac y as ce ntral. B eale r ca n be see n as a pur e Fr eg ean, ta king se riously Fr eg e' s ontology of conc epts wher e Peac ocke doe s not, since he offe rs a r eductive a ccount of conce pt possession that allows us to deemphasize the imp or ta nc e of the on tol og y of c on c e pts .

PAGE 44

35 Bealer I n th e fo llo wi ng , I wi ll g ive a br ie f c ha ra c te ri zat ion of B e a le r' s th e or y of c on c e pts , mai nl y as i t ap pear s i n "A T heo ry of C on cept s an d C on cept P os ses si on " (B eale r 1998b). As I sta te d a bo ve , B e a le r i s a F re g e a n a bo ut c on c e pts : My sta rt ing po int is t he tr uis m th a t the conc ept of being F is a c on c e pt. . . .I ho ld that conce pts, in thi s primary sense, a re su i ge ne ris irreduc ible entities comprising t h e o n t o l o gi ca l ca t ego ry i n t er m s o f w h i ch p ro p o s i t i o n s (t h o u gh t s , i n Fr ege 's sense) are to be ana ly zed. (B eale r 1998b, 261) Con se qu e ntl y , B e a le r t hin ks tha t it is m isl e a din g to r e du c e ta lk a bo ut c on c e pts to t a lk about lang uag e. Tha t is, if concepts a re mind-indepe ndent entities that form their ow n ontologica l ty pe, then they cer tainly cannot be identical with any linguistic items. There are many differ ent kinds of conc erns that a re a t play in the backg round of B eale r' s theory of conc epts, for e x ample, var ious kinds of realism, but for our pur poses we ne ed not conce rn ourse lves with those. Be aler say s expl icitly that he is a F reg ean a bout the on tol og ic a l st a tus of c on c e pts : I n the tradition of F reg e' s critique of psy cholog ism, my view is that propositions (a nd the c on c e pts in t e rm s o f w hic h th e y a re a na ly zab le ) a re on tol og ic a lly independe nt of the mind. Propositi ons are independe ntly require d for the purposes of log ical theor y , and they have the modal status one would expect logic al objects to have . (B eale r 1998b, 262) Since propositions are a naly zable in terms of conc epts, conc epts, too, have the modal status of log ical objec ts. As we ca n see, B eale r ac cepts F reg e' s third rea lm of objects, or the abstra ctness of c oncepts. Althoug h sever al philosophers ca ll themselves neoFr eg eans, not many of them hold on to the ontologic al commitments of Fr eg e' s theory of conce pts. Bea ler a lso states, "c oncepts may be def ined as those nonc ontinge nt entities which play the role of predic ate e ntities in t he log ical ana ly sis of predica tive and g ener al fi ne-grai ned pro po si ti on s" (B eale r 1998b, 267). F urthermor e, " [c] oncepts a nd

PAGE 45

36 propositions function as bea rer s of truth, log ical truth, nec essity , etc. a nd also as cogni ti ve an d l in guis ti c con ten ts " (B e a le r 1 99 8b , 2 68 ). As c on te nts of lin g uis tic ite ms, conce pts, then, cannot be reduc ed to the ling uistic items themselves. As cog nitive contents, we can se e B eale r' s use of ' conce pt' a s along the same lines a s my use of the te rm —a c on sti tue nt o f a tho ug ht c on te nt. I n his theory of conc epts, Be aler is primarily interested in g iving a n acc ount of conce pt possession because it seems that he take s what he thinks conc epts are for g ra nte d— on tol og ic a l e nti tie s th a t a re ind e pe nd e nt f ro m th e min d. F ur the rm or e , c on c e pts "wou ld ex is t w het her o r no t t hey app ly to any th in g" (B eale r 1998b, 262). F or example, there is a conc ept of a un ic or n and a c oncept of a ro un d s qu ar e even thoug h they do not apply to any thing in the wor ld. That is to say , the conc ept of a conce pt do e s n ot r e a lly ne e d to be a na ly zed a ny fu rt he r t ha n h e a lr e a dy ha s a na ly zed it b e c a us e he ta ke s c on c e pts to b e pr imi tiv e on tol og ic a l e nti tie s. Th e int e re sti ng qu e sti on fo r h im, the n, is w ha t it is t o possess those mind-independe nt entities, which are the contents of thoug hts. I n g iving his theory of conc ept possession, Be aler (1987, 1992, 1998b, 2002a) a ssu me s th a t pr op os iti on a l a tti tud e s a re re a l a nd tha t on e kin d o f p ro po sit ion a l a tti tud e is the starting point of g iving a n acc ount of conc ept possession. Be aler explains concept po sse ssi on in t e rm s o f i ntu iti on s: " I ho ld t ha t c on c e pt p os se ssi on is t o b e a na ly zed in terms of a cer tain kind of reliable pattern in one ' s intuit ions. The cha lleng e is to find an anal y si s t hat is at o nce n on cir cul ar an d fu ll y general " (B eale r 1998b, 262). I t seems that Be aler appea ls to int uitions i n his analy sis of conce pt possession because he take s the first-per son acc ount of conc ept possession as primar y (i.e., we should assume that we have a conce pt and then g o on to analy ze it). I t seems that he take s the first-pe rson acc ount as primary beca use he doe s not think we have a ny thing e lse to g o on.

PAGE 46

37 I sh ou ld p oin t ou t th a t mo st o f t his ta lk a bo ut i ntu iti on s is int ro du c e d in his 25 a rg ume nts a g a ins t e mpi ri c ism . Se e a lso 26 B e a le r 1 98 7, 19 96 fo r m or e dis c us sio n a bo ut i ntu iti on s. B e fo re we g o in to a ny de ta il a bo ut h ow B e a le r t hin ks int uit ion s p la y a ro le in c on c e pt p os se ssi on , w e sh ou ld g e t c le a r o n w ha t in tui tio ns a re . A n in tui tio n, fo r B e a le r i s a se e min g : For y ou to have a n intuiti on that A is just for it to seem to y ou that A. Her e ' se e ms' is u nd e rs too d, no t a s a c a uti on a ry or " he dg ing " te rm , b ut i n it s u se a s a te rm fo r a g e nu ine kin d o f c on sc iou s e pis od e .. thi s k ind of se e min g is intellectual , not sensory or introspec tive (or imag inative). (B eale r 1998b, 271) An intuition i s a propositional attitude that is disti nct from a belief, since one ca n have a n intuiti on without having the corr esponding belief. F urthermor e, B eale r states, " when we speak of intuiti on, we mea n a p rio ri in tu it io n" (B eale r 1992, 102). An a p rio ri intuiti on is goin g to "p res ent it sel f as neces sar y " (B eale r 1992, 102). For example, I may have 25 no intuiti ons about some mathematica l theorem be cause I have ne ver done the proof f or it, bu t I ma y sti ll b e lie ve tha t it is n e c e ssa ri ly tr ue . I t se e ms, the n, tha t w ha t B e a le r i s g oing to sa y about the re lation betwee n conce pt possession and intuiti ons is something along the following lines: someone possesses a conce pt, C , insofar a s she ca n have the corr ect intuitions about whether C a pp lie s. F or B e a le r, int uit ion s f un c tio n in the sa me way in justi fy ing our beliefs a s other experienc es do, but intuiti ons do a better job: "the standard justifica tory proce dure c ounts, not only experience s, observa tions, memory , and te sti mon y a s prima facie evi den ce, b ut in tu it io ns as w ell " (B e a le r 1 99 2, 10 1) . Th a t is , 26 int uit ion s a re a t le a st o n e qu a l f oo tin g wi th o the r w a y s w e fo rm be lie fs . We sh ou ld n ote that Be aler thinks that int uitions are fa llible: "in say ing tha t the standard pr ocedur e c ou nts int uit ion s a s pr im a fa c ie evidenc e, we do not prec lude using the mecha nism of

PAGE 47

38 self-c riticism to elimi nate intuition as a sourc e of prima facie bel ief " (B eale r 1992, 101). Fur thermore , I ntuitions play a sig nificant role in our beliefformation proc esses. F irst, at any g iven time, there are a number of novel que stions about which one has no be lief o n e w a y o r t h e o t h e r b u t a b o u t w h i c h o n e w o u l d h a v e a c l e a r c u t i n t u i t i o n . In case s like this, one will ty pically form the be lief assoc iated with the intuition as soon as the intuition occurs. Second, intuition play s a cr ucial role in following rules and pr ocedur es—for example, rules of infe renc e. (B eale r 1992, 104) Although intuitions are f allible, there are three important fea tures of intuitions tha t B e a le r t hin ks ma ke s th e m a re lia ble wa y of a c qu ir ing be lie fs : First, a pe rson' s intuit ions are la rg ely consistent with one another . To be sure , a g iven per son' s intuit ions occa sionally appea r to be c ontradictory , but so do our observa tions, our memories, and e ven our pur e sense experience s. . . . Second, a lth ou g h d if fe re nt p e op le do ha ve c on fl ic tin g int uit ion s f ro m ti me to t ime , th e re is a n im pr e ssi ve c or ro bo ra tio n b y oth e rs of on e ' s e le me nta ry log ic a l, m a the ma tic a l, conce ptual, and modal intuitions. . . . Third . . . int uition is seldom, if ever , disconfirmed by our experienc es and obse rvations. The pr imary rea son is that the contents of our intuiti ons—whether conce ptual, logic al, mathematica l, or modal—are by and lar g e indepe ndent of the c ontents of our obser vations and experience s. (B eale r 1992, 110) Be aler arg ues that many of the a ppare nt inconsistencies we have in our intuiti ons are the result of not having all of the re levant informa tion we nee d to fig ure out wha t our int uit ion s a re . F or e xamp le , b e fo re we he a rd a bo ut G e tti e r ( 19 63 ) c ou nte re xamp le s, mos t of us thoug ht that if we ha d a justified, true be lief, we had knowledg e. Howe ver, onc e we a re int ro du c e d to a Ge tti e r c ou nte re xamp le we ha ve dif fe re nt i ntu iti on s a bo ut w ha t is sufficie nt for knowledg e. The re is no conf lict in our intuit ions; before we we re introduced to Ge ttier counter examples, we just did not consider those kinds of ca ses and made a hasty g ener alization. Until we did conside r them, none of our intuitions were overturne d. Be aler claims that there are two senses in which one can posse ss a conc ept, a ' nominal' se nse and a ' full' sense. The following is how Be aler char acte rizes nominal

PAGE 48

39 See 27 Bur g e 1979. conce pt possession: "A subjec t possesses a g iven conc ept at lea st nominally iff the subject has na tural propositional attitudes (belief , desire, e tc.) towa rd propositions which hav e th at c on cept as a c on cept ual con ten t" (B eale r 1998b, 272). Nominal c oncept po sse ssi on is s up po se d to ma ke ro om f or po sse ssi ng c on c e pts tha t w e do no t c omp le te ly understand. F or example, I nominally possess the conc ept of ar thr iti s if I have be liefs a bo ut a rt hr iti s, bu t I so me tim e s mi sa pp ly it. No min a l c on c e pt p os se ssi on is u se fu l in 27 third person a ttributions of concept possession. That is, we can a ssume that when a n individual uses a term that e x presse s a conc ept, he or she at lea st has nominal possession of the c oncept e x presse d by the term (i.e ., we c an assume tha t even if the individual has many g aps in his or her unde rstanding of the c oncept, she can a t least have beliefs that ha s th e c on c e pt a s p a rt of its c on te nt) . T he re a re c a se s in wh ic h s ome on e a pp a re ntl y p o s s es s es a c o n ce p t , b u t i s i n ce rt ai n wa y s d ef i ci en t ; m ay b e t h i s i s n o t re al l y p o s s es s i n g a conce pt at all. Although this is an intere sting issue, it is not important to di scuss it here. B e a le r c ha ra c te ri zes the str on g , o r f ull , s e ns e of c on c e pt p os se ssi on a s f oll ow s: A s ub je c t po sse sse s a c on c e pt i n th e fu ll s e ns e if f ( i) the su bje c t a t le a st n omi na lly po sse sse s th e c on c e pt a nd (i i) the su bje c t do e s not do this with m isunderstanding or incomplete unde rstanding or just by virtue of sa tisfy ing our attribution prac tices or in any other we ak such w ay . (B eale r 1998b, 273) Given condition (ii), it seems that to fully possess a conc ept one must not have a ny misunderstanding about it. Howeve r, it is unclear what B eale r thinks it is to have a complete unde rstanding of a c oncept. B eale r thinks that when we talk about someone un de rs ta nd ing a c on c e pt w e me a n th a t th e ind ivi du a l f ull y po sse sse s th e c on c e pt. F ull conce pt possession is what is useful in first-person a ttributions of concept possession.

PAGE 49

40 I n 28 (B eale r 2002a) , Be aler talks about deter minate under standing of a c oncept rathe r than de terminate c oncept possession, but they seem to come to the same thing . I n " Mo da l E pis te mol og y a nd Ra tio na lis m," B e a le r g ive s a mor e pr e c ise 29 e xpla na tio n o f w ha t he me a ns by de te rm ina te un de rs ta nd ing of a c on c e pt: [D]e te rm ina te un de rs ta nd ing = the mod e m of un de rs ta nd ing su c h th a t, nece ssarily , for a ll x and all p under stood m-ly by x , (a) p is true if it i s p os sib le for x to settl e with a pr iori stability that p is true (b.i) p is true on ly if it i s p os sib le for x to settl e with a pr iori stability that p has a c ounterpa rt that is true (f or proper ty identities p) (b.ii) p is true on ly if it is poss ible for x to believe m-ly that p is true (for p believable by x ). (B eale r 2002a, 106) To put what B eale r has in mind more simply , condition (a) is one acc ording to which the individual who determinate ly possesses the c oncept a lway s intuit s the same thing about the conc ept g iven her deter minate under standing of the c oncept. Condition (b.i) is one Now that we have se en that B eale r (1987, 1992, 1998b, 2002a ) is interested in the no tio n o f f ull c on c e pt p os se ssi on in h is a na ly sis , w e ne e d to se e ho w i ntu iti on s p la y a ro le in f ull c on c e pt p os se ssi on . A lth ou g h ' int uit ion ' do e s n ot a pp e a r i n h is a na ly sis of fu ll c on c e pt p os se ssi on , h e do e s th ink tha t a n in tui tio n is a pr op os iti on a l a tti tud e , a nd the fi rs t condition of nominal conce pt possession talks about having propositional attitudes. Be aler explains t he role of intuitions i n deter minate, or full, conc ept possession as fo llo ws : Deter minate conc ept possession might be e x plicated ( at least in par t) in terms of the metaphy sical possibilit y of truth-tra cking intuiti ons in appropria tely g ood c og nit ive c on dit ion s. . . . w he n a su bje c t' s mo de of c on c e pt p os se ssi on sh if ts t o de te rm ina te ne ss t he re is a n a sso c ia te d s hif t in the po ssi ble int uit ion s a c c e pta ble to the su bje c t. I n f a c t, t he re is a sh if t bo th i n q ua nti ty a nd qu a lit y . T he qu a nti ty g ro ws be c a us e inc omp le te un de rs ta nd ing is r e pla c e d w ith c omp le te un de rs ta nd ing , e lim ina tin g " do n' t kn ow s" . T he qu a lit y imp ro ve s b e c a us e incorre ct under standing is replac ed with cor rec t understanding . (B eale r 1998b, 279) 28 When we move f rom nominal possession of a conc ept to deter minate possession of a conce pt, there is a shift in the individual' s intuit ions—the individual has more intuitions (q u an t i t y ) a n d b et t er o n es (q u al i t y ) ( i . e. , o n es t h at ac t u al l y t ra ck t h e t ru t h ). W h en "d o n 't kn ow s" a re e lim ina te d, the re su lt i s a g re a te r c on fi de nc e in o ur int uit ion s. 29

PAGE 50

41 acc ording to which an individual meets (a ) for Twin Ear th case s. Conditi on (b.ii) is one a c c or din g to w hic h a n in div idu a l be lie ve s th a t p i s tr ue wh ile sti ll u nd e rs ta nd ing it i n mode m (B eale r 2002a, 106) . I n this section, I have e x plained that B eale r thinks that Fre g e wa s rig ht when he introduced c oncepts a s abstrac t entities. Si nce he takes F reg e' s picture f or g rante d, he moves on to g ive an a ccount of what it is to possess a conce pt. Given the ana ly sis that Be aler g ives of conc ept possession, we c an see that he see ms to thi nk that for e very situation, it is determinate w hether or not a c oncept a pplies. I n the following section, I wi ll e xpla in P e a c oc ke ' s th e or y of c on c e pts . P e ac oc ke I n th is s e c tio n, I wi ll e xpla in a no the r c on te mpo ra ry the or y of c on c e pts , Peacoc ke' s, which also pay s car eful a ttention to the notion of gra sping, or possessing, a conce pt. I will primarily sketch Pea cocke ' s theory of conc epts as it appea rs in "What Ar e Concepts? " a nd A S tud y of C on c e pts (P ea co ck e 1 9 8 9 , 1 9 9 2 ). Al t h o u gh P ea co ck e's theory of conc epts is quite detailed, I will only explain thos e fe ature s of his theory that a re us e fu l f or my pu rp os e s. Pe a c oc ke e xpla ins his us e of ' c on c e pt' a s f oll ow s: The c oncepts whic h are my intended subjec t-matter a re ve ry roug hly indicated by say ing tha t they can be taken a s Fre g ean se nses. They may be of a ny cate g ory —singula r, pre dicative, or of hig her le vel. Withi n any one ca teg ory , they may be of va rious ty pes: descr iptive or demonstrative in the sing ular c ase; rela tively observa tional or hig hly theore tical in the pre dicative se nse. (Peac ocke 1989, 1) As we c an see , this chara cter ization di ffe rs slightly from the one I prese nted ea rlier in the c ha pte r a c c or din g to w hic h c on c e pts a re on ly the se ns e s o f p re dic a te s. Th a t is , w he re my no tio n o f c on c e pt o nly a llo ws fo r s e ns e s o f p re dic a te s, Pe a c oc ke ' s a lso a llo ws c on c e pts to be the se ns e s o f s ing ula r t e rm s a nd hig he r l e ve l te rm s. F ur the rm or e , h e sta te s,

PAGE 51

42 My conce pts, even in the c ase of a first-le vel pre dicative c oncept, a re thus distinct from wha t are called " conce pts" in Eng lish translations of Fre g e, i.e., f unctions from objec ts to truth values. My conce pts are a lso to be clea rly distinguished f r o m p r o p e r t i e s . T h e r e c a n b e m a n y d i f f e r e n t c o n c e p t s o f t h e s a m e p r o p e r t y. (Peac ocke 1992, 2) B oth Pe a c oc ke a nd I us e ' c on c e pt' dif fe re ntl y tha n F re g e us e s ' Be gr iff ' . The distinction that Peacoc ke wa nts to draw betwe en conc epts and prope rties is that proper ties are the re fe re nts of so me e xpre ssi on s, wh e re c on c e pts a re no t. T he re su lt o f c on fu sin g the tw o is simil ar to wha t happens whe n "c onfusing theories of modes of pre sentation with theories abo ut th e ob jec ts so pres ent ed" (Peac ocke 1992, 2) . There is a kind of level c onfusion. Pe a c oc ke e xplic itl y sta te s w ha t a bo ut F re g e ' s n oti on of se ns e is i mpo rt a nt i n ind ivi du a tin g c on c e pts : Th e str a nd in F re g e ' s c on c e pti on of se ns e wh ic h ma tte rs to m e he re is t ha t in which distinctions between se nses ar e tied to diffe renc es in potential informativene ss. One sense or conc ept is distinct from a sec ond if substitut ion of the first for the sec ond in a Thoug ht can y ield a Thoug ht with a differ ent cog nitive significa nce. (Peac ocke 1989, 1) This is also something that I stated ea rlier in the se ction entitled, "Star ting Points and Terminolog ical I ssues" : two thoughts diffe r fr om one another just in case either (a) the tho ug hts c on ta in a t le a st o ne dif fe re nt c on c e pt, (b ) t he tho ug hts c on ta in t he sa me c on c e pts bu t th e ir mod e of c omb ina tio n is dif fe re nt, or (c ) b oth (a ) a nd (b ). We s ho uld note that Peac ocke a g ree s with Bea ler a bout whether the possession of conc epts requir es a n a bil ity to s pe a k a la ng ua g e : " the g e ne ra l f ra me wo rk of po sse ssi on c on dit ion s is its e lf neutra l on the issue of whe ther some, a ll, or no conce pts require lang uag e for mastery for th eir po ss ess io n" (Peac ocke 1992, 31) . Whe re a s B e a le r' s th e or y of c on c e pts be g ins wi th a g e ne ra l a na ly sis of wh a t it is t o de te rm ina te ly po sse ss a ny c on c e pt, Pe a c oc ke ' s th e or y of c on c e pts is g ro un de d in answe ring the following question: "What form should be take n by a theor y of a pa rticular

PAGE 52

43 con cept ? " ( P e a c o c k e 1 9 8 9 , 2 ) . T h e f o r m t h a t P e a c o c k e t h i n k s s h o u l d b e t a k e n b y a the or y of a pa rt ic ula r c on c e pt i s th e fo llo wi ng : " Th e c on c e pt F is t ha t c on c e pt C to possess which a thinker must meet the condition A ( C )" (Peac ocke 1989, 2) . Peacoc ke agr ee s wi t h Be al er ab o u t a c en t ra l n o t i o n we s h o u l d b e c o n ce rn ed ab o u t wh en gi v i n g a the or y of c on c e pts , n a me ly , th e no tio n o f c on c e pt p os se ssi on , o r g ra sp ing a c on c e pt: I n this material, I will try to respec t the following Principle of De pendenc e: the pr inc ipl e tha t th e re c a n b e no thi ng mor e to t he na tur e of a c on c e pt t ha n is de te rm ine d b y a c or re c t a c c ou nt o f t he c a pa c ity to h a ve pr op os iti on a l a tti tud e s to contents conta ining that c oncept ( a cor rec t acc ount of " g rasping the conc ept" ). (Peac ocke 1989, 2) Pos se ssi on c on dit ion s a re a n in div idu a l' s r e a so ns fo r f or min g be lie fs (Peac ocke 1992, 55 ). Ho we ve r, " Tw o th ink e rs wi th t he sa me c on c e ptu a l r e pe rt oir e a nd in t he sa me evidential position may differ in which contents they are disposed to acc ept, bec ause one is bo ld er, m ore i magi nat iv e, o r m ore i ngeni ou s t han th e ot her" ( Pe a c o c k e 1 9 8 9 , 2 3 ) . We wi ll s e e be low tha t c on c e pt a ttr ibu tio n c on dit ion s a re we a ke r t ha n p os se ssi on c on dit ion s, tha t is , e ve n if ind ivi du a ls m a ke ma ny mis ta ke s b a se d o n s ome mis un de rs ta nd ing in a pp ly ing a pa rt ic ula r c on c e pt, we ma y sti ll a ttr ibu te to t he m th e c on c e pt. Where B eale r (1987, 1992, 1998b, 2002a ) is conce rned w ith giving a theor y that is v e ry g e ne ra l, P e a c oc ke is c on c e rn e d w ith g ivi ng a mor e pa rt ic ula r t he or y . T ha t is to say , wher e B eale r g ives a theor y of wha t it is to possess any conce pt, Peacoc ke g ives an analy sis of what it is to possess differe nt conce pts by introducing the form that e very a c c ou nt m us t f oll ow , w hic h in c lud e s w ha t th e ind ivi du a l mu st k no w i n o rd e r t o p os se ss the re le va nt c on c e pt. I n g e ne ra l, t ho ug h, Pe a c oc ke sta te s th a t " po sse ssi ng a c on c e pt i s knowing what it is for something to be its semantic value" and that prope rties ar e " the sem ant ic v alu es o f pr edi cat iv e con cept s" (Peac ocke 1992, 23) . He a lso states, "I f the se ma nti c va lue of a c on c e pt i s a pr op e rt y , th e n . . . g ra sp of a c on c e pt, me e tin g its

PAGE 53

44 possession condition, is knowledge of wha t it is for a proper ty to be the sema ntic value of th e con cept " (Peac ocke 1992, 23) . Fur thermore , he states, " The possession conditions wi ll c a ptu re on ly wh a t ve ry min ima l r a tio na lit y is r e qu ir e d f or g ra sp of the c on c e pts in qu est io n" (Peac ocke 1992, 39) . Peacoc ke ar g ues that the ne cessa ry conditions for possessing a conc ept, C , mu st not be circ ular (i.e ., they must not presuppose that the individual alre ady possesses the conce pt). Under standing the nec essar y conditions for possessing a conc ept, C , must not re qu ir e the ind ivi du a l to be a ble to h a ve pr op os iti on a l a tti tud e s w ho se c on te nt c on ta ins C . F or e xamp le : Th e c on c e pt red is t ha t c on c e pt C to possess which a thinker must: (i) be disposed to judge that a per ceptua lly g iven object f alls under C when it is prese nted in a ' red' reg ion of his visual field in conditions he takes to be nor mal and whe n he takes his per ceptua l mechanisms to be wor king pr operly , and to make the judgme nt for the r eason tha t it is so prese nted; and (ii) the thinker must be disposed to judge of an obje ct not so prese nted per ceptua lly that it falls under C when he takes it to have the primary quality g round (if a ny ) of the disposition of objects to ca use experienc es of the sor t mentioned in (i). (Peac ocke 1989, 3) Simi lar to other the ories I have e x plained, conc ept possession is closely rela ted to having cer tain dispositi ons in certa in situations. As we can se e, the ne cessa ry conditions for possessing the conce pt of red do not contain ' red' in a content c lause of a propositional attitude. Peacoc ke thinks that there is a substantive differ ence betwee n possession c on dit ion s a nd a ttr ibu tio n c on dit ion s: A possession condition states what is requir ed for full mastery of a pa rticular conce pt. The attribution conditions for red , the conditions under whic h so me thi ng of the fo rm " x believes tha t ___red___" is true, ar e much we aker than the possession condition. (Peac ocke 1992, 29) That is to say , we a llow that someone will attribute a c oncept to an individual eve n if he or she make s some conce ptual err ors, but do not say that such an individual possesses the conce pt. For Pea cocke , possession conditions hold across possible worlds, (e .g ., the

PAGE 54

45 possession conditions for the conc ept of red a re the sa me no ma tte r w ha t w or ld o ne is in): Even if a possession condition is st ated e ntirely nonmodally , when it is put forwa rd as a possession condition, it s hould be under stood as incurring the c omm itm e nt t ha t in wo rl ds oth e r t ha n th e a c tua l w or ld, the re is n oth ing mor e to being the conc ept in question than having the same posse ssion condition as has been sta ted for the actua l world. (Peac ocke 1992, 47) The ne x t step in Peacoc ke' s theory of conc epts is to show "how the semantic va lue of the conce pt is determined fr om its poss ession conditions (toge ther with the wor ld)" (Peac ocke 1989, 6) . He state s, "Me eting the possession condition for a c oncept c an be ide nti fi e d w ith kn ow ing wh a t it is f or so me thi ng to b e the c on c e pt' s se ma nti c va lue (i ts refere nce) " (Peac ocke 1989, 8) . That is to say , possessing a conce pt is just knowing the conditions under which some thing is the re fer ent of a conce pt. I n other wor ds, conce pt possession is knowing wha t the contribution of a c oncept is to the truth conditions of the proposition in which it appear s. The contr ibution of the conce pt to the truth conditions of a pr op os iti on in w hic h th e c on c e pt a pp e a rs is t he se ma nti c va lue of the c on c e pt. I wi ll no t g o in to d e ta il o n th is a sp e c t of Pe a c oc ke ' s th e or y , h ow e ve r, sin c e it i s n ot d ir e c tly rele vant to what I am intere sted in here . I nstead, we will ge t clear on the extent to which Pe a c oc ke thi nk s th a t c on c e pts a re a bs tr a c t ob je c ts. Peacoc ke notes the pr oblem of making sense of how abstra ct objects c an be a part of the menta l states of individuals as one that he must address in his theory of conc epts. To solve this problem, Peacoc ke does not re duce c oncepts to something else, but he doe s reduc e the possession conditions for c oncepts, ther eby making it unproblematic for a bs tr a c t ob je c ts t o b e pa rt of a de sc ri pti on of the ph y sic a l w or ld. He re du c e s e pis te mic conce pts to non-epistemic, that is, phy sical, conc epts. Peac ocke doe s this by introducing

PAGE 55

46 the notion of primitive compellingne ss about cer tain infere nces w e make . I n particula r, he dis c us se s th e pr imi tiv e c omp e lli ng ne ss o f i nf e re nc e s in vo lvi ng log ic a l c on c e pts : Con jun c tio n is tha t c on c e pt C to possess which a thinker must find transitions t hat a re ins ta nc e s o f t he fo llo wi ng fo rm s p ri mit ive ly c omp e lli ng , a nd mus t do so be c a us e the y a re of the se fo rm s: p q pCq pCq ___ ___ ___ pCq p q To sa y tha t th e thi nk e r f ind s su c h tr a ns iti on s p ri mit ive ly c omp e lli ng is t o s a y thi s: (1) he finds them compelling ; (2) he doe s not find them compelling be cause he has infer red the m from other pr emises and/or pr inciples; and (3) for possession of the c on c e pt C in question (here conjunction) he does not nee d to take the corr ectne ss of the transitions as answe rable to any thing e lse. (Peac ocke 1992, 6) Th e no tio n o f p ri mit ive c omp e lli ng ne ss p la y s th e sa me ro le a s B e a le r' s in tui tio ns do in his the or y of c on c e pts . Wh e re B e a le r ( 19 87 , 1 99 2, 19 98 b, 20 02 a ) t hin ks tha t a c e rt a in reliability of our intuitions is what is i mportant to conce pt possession, Peacocke thinks tha t f ind ing c e rt a in i nf e re nc e s p ri mit ive ly c omp e lli ng is w ha t is imp or ta nt: A possession condition for a pa rticular c oncept spe cifies a role that individuates tha t c on c e pt. Th e po sse ssi on c on dit ion wi ll m e nti on the ro le of the c on c e pt i n cer tain transitions that the thinker is willing to make. The se will be transitions that involve complete pr opositional thoughts involving the c oncept. (Peac ocke 1992, 107) One main diff ere nce be tween B eale r and Pea cocke is that Bea ler g ives a non-r eductive ac co u n t o f c o n ce p t p o s s es s i o n wh er e P ea co ck e gi v es a r ed u ct i v e a cc o u n t . P ea co ck e's a c c ou nt i s r e du c tiv e be c a us e the pr imi tiv e ly c omp e lli ng inf e re nc e s th a t he a pp e a ls t o in e xpla ini ng c on c e pt p os se ssi on a re ps y c ho log ic a l st a te s th a t a re su pp os e d to be re du c ibl e to n a tur a l, t ha t is , n on -e pis te mic , s ta te s. Sin c e Pe a c oc ke (1 98 9, 19 92 ) t a lks a bo ut c on c e pts pr ima ri ly in t e rm s o f t he ir possession conditions, a conce pt's posse ssion conditions are stated in ter ms of pr op os iti on a l a tti tud e s, a nd pr op os iti on a l a tti tud e s a re su pp os e d to be re du c ibl e to

PAGE 56

47 phy sical states, he is able to dee mphasize the importance of abstrac tness as a f eatur e of c on c e pts . Co nc e pts the mse lve s a re no t li te ra lly pa rt of the me nta l st a te s o f i nd ivi du a ls beca use we talk about conc epts in terms of their posse ssion conditions. P ossession conditions can be reduc ed to non-e pistemic (that is, phy sical) c onditions, and since c on c e pt p os se ssi on is c e ntr a l to Pe a c oc ke ' s d isc us sio n o f c on c e pts ra the r t ha n c on c e pts themselves, Peac ocke c an make room for c oncepts be ing e mploy ed in a phy sical descr iption of the world. Whether Pea cocke ' s story about this is plausibl e or not is an interesting issue, but it is not one that I will address he re. Taki ng Stock I n this chapter, I sketche d six theories of conce pts in an attempt to ge t clear on how diffe rent philosophers use the term ' conce pt'. T he theor ies I sketche d differ from one anothe r in a va riety of wa y s (e.g ., they g ive diffe rent a ccounts of the fea tures of conce pts, what it is to pos sess a c oncept, a nd they use the ter m 'c oncept' in differ ent way s). I n the theorie s I discussed, I tried to make c lear how the theor y connec ts with the way I talk about conc epts in this dis sertation. I n some ca ses, it turns out that the philosopher is using ' conce pt' in quite a differ ent way than I do, and in other c ases, it took so me c los e loo kin g to s e e wh e the r ' c on c e pt' is u se d in the wa y I us e it. We a re no w a ble to s e e ho w w e c a n ma ke se ns e of the c la im t ha t c on c e pts a re c on sti tue nts of tho ug hts , which fits all of the the ories discussed in this chapte r. So, this seems to be a cor e fe ature of c on c e pts a nd so me thi ng in v ir tue of wh ic h w e c a n s a y we a re ta lki ng a bo ut t he sa me subject. We now have a wa y of identify ing the subject matter. There are cer tain fea tures of c oncepts that will be c entra l to the subsequent discussions in thi s dissertation, and I will discuss them in detail in t he following two c ha pte rs : c on c e pt p os se ssi on , c on c e pt i nd ivi du a tio n, a nd wh a t c on c e pts a re . I wi ll s pe ll

PAGE 57

48 out the notion of conce pt possession in terms of epistemic conditions, that is, in terms of kn ow ing wh e n to ma ke c e rt a in j ud g me nts a bo ut w he the r a c on c e pt a pp lie s, a nd I wi ll elabor ate it to handle na tural kind conc epts. I n doing this, I will ex plain how things a re in t he a c tua l w or ld i s r e le va nt t o c on c e pt p os se ssi on . I n tr a dit ion a l th e or ie s o f c on c e pts , nothing a bout the actua l world is releva nt to concept possession. Althoug h I do not explain how each tr aditional theory of conc epts I discussed in this chapter individuates c on c e pts , w e c a n s a y tha t on the se vie ws a c on c e pt i s in div idu a te d b y the pr op e rt y it picks out (e.g ., on Wit tge nstein' s view, we can sa y that two conce pts are ide ntical iff they have the proper ty of re sembling some pr ototy pe in some wa y ). On my view conc epts are associate d with two differ ent prope rties, so they must be individuated differ ently . The fea tures of c oncepts that will not be ce ntral in subsequent discussions are de te rm ina te ne ss a nd a bs tr a c tne ss. Abstract Theories Ver sus Use Theories I have a lrea dy discussed at leng th the idea that c oncepts a re a bstract, so I will not dis c us s it in d e ta il h e re . Wh a t w e ha ve se e n in the the or ie s I sk e tc he d in thi s c ha pte r i s that only Fr eg e and B eale r think that conce pts are a bstract objec ts in a strong sense. Wit tge nstein reduc es talk about c oncepts to talk about the w ay we use words, whe rea s Gea ch talks about c oncepts a s the meaning s of words ( not by just adopting a view of F re g e ' s, bu t in te rm s o f u se a s Ry le ' s o r Wi ttg e ns te in' s) , a nd Ry le ta lks a bo ut c on c e pts in terms of dispositions, or capa cities, to make c erta in judgments. B eale r and Pea cocke emphasize the importance of conc ept possession in their theories of conce pts, but where Be aler is a pure F reg ean, Pea cocke reduc es talk about c oncepts to talk about c oncept possession, which ca n be re duced to talk a bout phy sical conditions to avoid a commitment to su i ge ne ris entities full-stop. I t is im portant to note the re lation betwee n

PAGE 58

49 Wit tge nstein uses the term ' g ame' beca use it expresses a fa mily rese mblance 30 c on c e pt. Con c e pts by the ir na tur e do no t ha ve ru le s la id o ut i n a dv a nc e fo r a ll o f t he ir application. Although the application of the rules is not determinate , the rules themse lves 31 might be. c on c e pts a nd la ng ua g e , th a t is , p re dic a te s a nd oth e r g ra mma tic a l e xpre ssi on s, wh ic h mo st of the philosophers I discussed empha siz e in their theor ies of conc epts. Vagueness and Con cepts I a dd re sse d th e iss ue of va g ue pr e dic a te s in se ve ra l th e or ie s o f c on c e pts tha t I discussed in this chapter . The issue of va g ue pre dicates is close ly rela ted to the issue of deter minateness of c oncepts. I f conc epts are not by nature deter minate, that is, if there are at least some c ases in which the conce pt does not g ive us ade quate g uidance about whether it applies or fa ils to apply , then we mig ht say that some conc epts are vag ue. Of the philosophers I discussed in this chapter , the only one who a rg uably thinks that predic ates a re va g ue by na tur e i s W i t t gen s t ei n , wh o em p h as i z es t h at al l u s e o f l an gu age is a matter of play ing la ng uag e-g ames, the rules of which do not fully deter mine 30 application conditions for the predic ates. That is, there are cer tain situations where our 31 r u l e s f o r u s i n g p r e d i c a t e s d o n o t t e l l u s w h e t h e r t h e p r e d i c a t e a p p l i e s o r f a i l s t o a p p l y, and ther e is no rea son to think t hat the use of a wor d must determine for all conditions wh e the r i t is to a pp ly or no t. F re g e , o n th e oth e r h a nd , a rg ue s th a t a ny pr e dic a te tha t is vag ue does not e x press a conce pt at all. Ry le, althoug h he does not a ddress the issue of va g ue ne ss d ir e c tly , w ou ld s e e m to a rg ue tha t va g ue pr e dic a te s d o n ot e xpre ss c on c e pts , either. G eac h is silent on whether va g ue pre dicates e x press c oncepts. B eale r and Peacoc ke, since they are neo-F reg eans a bout conce pts, would seem to arg ue the wa y Fr eg e does.

PAGE 59

50 Whether vag ue pre dicates e x press c oncepts is important bec ause the set of our conce pts is differe nt depending on whether conce pts are the sort of thing s that can be vag ue. B y deny ing va g ue pre dicates e x press c oncepts a t all, some philosophers have restric ted themselves to a f ar smaller set of conc epts than those philosophers who a llow c on c e pts to b e va g ue . A s a ma tte r o f f a c t, m os t of ou r p re dic a te s se e m to be va g ue , s o if vag ue pre dicates do not e x press c oncepts, then ma ny of the pre dicates w e mig ht have tho ug ht e xpre sse d c on c e pts do no t. F or e xamp le , th e pr e dic a te s ' is a pe rs on ' , ' is b a ld' , ' is a ca t', e tc. do not express conce pts. Predicates tha t do express conce pts on the view a c c or din g to w hic h o nly no nva g ue pr e dic a te s e xpre ss c on c e pts a re ' is a nu mbe r' a nd ' is a set' , and the like. Ot he r Iss ue s Th e re a re fu rt he r i nte re sti ng qu e sti on s a bo ut w ha t c on c e pts a re , in a dd iti on to whether they are abstrac t objects or not. Fr eg e, of c ourse, thinks that conc epts are literally a diffe rent kind of thing than any thing e lse (i.e., they are sui ge ne ris ). Other philosophers, such as Wittge nstein, arg ue that conc epts are not things in any sense of the term ' thing' , but, rather , talk about conc epts is cashe d out in terms of how expressions are us e d. I n f a c t, i t is no t e ve n a pp ro pr ia te to a sk Wittg e ns te in w ha t c on c e pts ar e . Ry le, on the oth e r h a nd , b e g ins his ta lk a bo ut c on c e pts a rg uin g tha t ta lk a bo ut c on c e pts is misleading in the first place and should be re place d by talk about the mea nings of g ener al terms. L ater , howeve r, he ta lks about the conc ept of if , which is clea rly not a g ener al term. I n the end, thoug h, it seems that Ry le ac tually thinks a conce pt can be understood as the mea ning of a wor d and the mea ning of a wor d is how the word is used in a lang uag e. Althoug h this sounds a bit l ike Witt g enstein, Ry le thinks that the rules for using w ords in a lang uag e alwa y s tell us whether a conce pt applies or fa ils to apply .

PAGE 60

51 After Ry le (a lthough we can se e hints of this in W ittgenste in), the notion of a conce pt takes on a slig htly differ ent role. I nstead of f ocusing on what conc epts are , philosophers seem to be pr imarily conce rned w ith what it is t o possess a conc ept. Gea ch seems to beg in this t rend by talking a bout conce pts as capa cities we e x erc ise when w e ma ke jud g me nts . We c a n s til l sa y tha t he thi nk s th a t th e re is a c omm on e le me nt i n dis tin c t th ou g hts , b ut h e sp e nd s mu c h o f h is t ime ta lki ng a bo ut c a pa c iti e s to ma ke c e rt a in judgme nts. I t is unclear on G eac h' s view wha t kinds of capa cities conc epts are limit ed to, if any . As a re sult, many philosophers seem to think that quite a lot of ca pacities we never thought of as re lated to conc epts ac tually are . This raises intere sting issues in the philosophy of mind, specific ally in how we think about phenome nologic al conc epts. A number of philosopher s attempt to cash out phenomena l states in terms of re cog nitional conce pts, which are conce pts spelled out in terms of a c apac ity that does not seem to be the no tio n tr a dit ion a lly a sso c ia te d w ith c on c e pts . B ut, if " ph e no me no log ic a l c on c e pts " are not conce pts at all, then it is difficult how we ca n g ive a c onceptua l analy sis of pa in . I f t he y a re , th e n w e ha ve to g ive a n a c c ou nt o f c on c e pts tha t ma ke s r oo m f or the m. Since B eale r and Pea cocke are neo-F reg eans, the y take the F reg ean pic ture of what conc epts are as primitive and foc us most of their discussion about conce pts on the notion of conce pt possession. The cha ng e in foc us to talk about conce pt possession seems to have also fuele d the trend to a llow many more thing s fall under the conc ept of a conce pt than is permitted on an a ustere vie w on what c oncepts should be limited to (i.e., more than just the sense s of pre dicates) . All of this disagr eeme nt about the var ious aspects of c oncepts make s it very difficult to talk about conc epts in a wa y that is ag ree able to eve ry one. We ca n see the rela tion differe nt discussions of conce pts (that is, most agr ee tha t conce pts are

PAGE 61

52 constituents of thoug ht), but there se ems to be a w ide ra ng e of w hat counts as a conce pt a nd wh a t is the mos t im po rt a nt t hin g to t a lk a bo ut w he n ta lki ng a bo ut c on c e pts . T ha t is , just because most phil osophers a g ree that conce pts are c onstituents of thought does not me a n th a t th e y a g re e a bo ut w ha t c ou nts a s a c on sti tue nt o f t ho ug ht a nd wh e the r c on c e pts are just a cer tain subset of constituents of thoug ht. This has cause d discussions about conce pts to be unclea r to the point that it seems as if philosophers ar e talking about differ ent things w hen they claim to be talking about conc epts. The introduc tion of a posteriori nec essities by Kripke ( 1980) and Putnam (1975) has adde d another complication to theories of conce pts—how to acc ount for conc epts that are associate d with more than one proper ty . I will discuss thi s in detail in the next section. The purpose of the sec tions on theories of conc epts was to lay out a historical fra mework f or a the ory of conc epts as a ba ckg round for a discussion of the utility and su c c e ss o f c on c e ptu a l a na ly sis . T he re a re ma ny dif fe re nc e s in the or ie s o f c on c e pts despite some similarities among them, which may aff ect how dif fer ent philosophers dea l wi th i ssu e s p e rt a ini ng to c on c e pts . I t is imp or ta nt f or the pu rp os e s o f t his dis se rt a tio n to focus on the notion of c oncept possession as B eale r (1987, 1992, 1998b, 2002a ) and Peacoc ke (1989, 1992) do, since I spend a lot of time talking about the role of thoug ht experiments in doing philosophy , the succ ess of whic h depends on the rele vant individuals' possessing conce pts. Contem porary Com plications: A Post erior i Nece ssities I n this section, I will introduce brief ly a conte mporary complication to any traditional theory of conc epts, namely , the discover y of ce rtain fe ature s of natura l kind conce pts that seem to g ener ate a posteriori nec essities by Kripke ( 1980) and Putnam (1975). The acc ount of natura l kind concepts provide d by Kripke a nd Putnam is by now

PAGE 62

53 See, for example, 32 Cha lme rs 19 96 ; Jac ks on 19 98 a . Wit me r ( 20 01 ) a nd Ya blo (2 00 0) a lso us e tw odim e ns ion a l se ma nti c ta lk, tho ug h th e y a re no t ne c e ssa ri ly c omm itt e d to a tw odim e ns ion a l se ma nti c s. well known and popula r enoug h to be ca lled the standar d acc ount, and I acc ept the acc ount in this dissertation. Kripke a nd Putnam have similar views a bout natural kind conce pts, so I talk about the standa rd ac count as c oming f rom both Kripke a nd Putnam. There have be en seve ral philosophers who ha ve explained the standa rd ac count of na tural kind conce pts by appea ling to a twodimensional semantics, but I will ex plain it rather g ener ally in this section and focus more on how it presents a complication to traditional theories of conce pts. Although the re a re subtle diff ere nces be tween K ripke' s and 32 Putnam's view s and subtle diffe renc es among philosophers' explanations of twod i m e n s i o n a l s e m a n t i c s , t h o s e d i f f e r e n c e s a r e n o t r e l e v a n t t o m y p u r p o s e s h e r e . In Chapter 2, I will ex plain two-dimensional semantics a nd its relevanc e to the cur rent projec t in more detail. A natura l kind concept is a c oncept that is assoc iated with two diffe rent kinds of proper ties. First, there is a proper ty that the natura l kind concept a ctually picks out or ref ers to; one c annot know a pr iori that the conc ept re fer s to that property . Second, ther e is a no the r p ro pe rt y , a c lus te r o f s up e rf ic ia l f e a tur e s u se d c omm on ly to i de nti fy thi ng s in t h e a ct u al wo rl d t h at fa l l u n d er t h e c o n ce p t . M o re fo rm al l y , we ca n s ay t h e f o l l o wi n g: A c on c e pt, C, i s a na tur a l ki nd c on c e pt i ff : (1 ) T he re is a pr op e rt y , P, su c h th a t ( a ) C in f a c t pi c ks ou t P; a nd (b ) i t is impossibl e to know a pr iori that C picks out P ; and (2 ) T he re is a pr op e rt y , Q , s uc h th a t ( a ) C do e s n ot p ic k o ut Q ; bu t ( b) it i s possible to know a priori that the e x tensions of C and Q ar e identica l in the actual world. One f amous example of a na tural kind conc ept is the conc ept of water . The prope rty of 2 being H O is the proper ty that the conc ept of water actua lly picks out; we ca nnot know a 2 pr ior i th a t water pic ks ou t th e pr op e rt y of be ing H O. Wha t water pic ks ou t is

PAGE 63

54 I wi ll d isc us s in de ta il h ow Put na m a rr ive s a t hi s c on c lus ion via the Tw in E a rt h 33 thought e x periment in Chapter 3. At this point , I only want to introduce a posterior i ne c e ssi tie s a s a pr ob le m f or tr a dit ion a l th e or ie s o f c on c e pts . deter mined by what we know a pr iori (whate ver w ater is, it has the property of being the wa te ry stu ff a ro un d h e re ) p lus the wa y thi ng s a re in t he a c tua l w or ld. We c a n c ome to know a pr iori that the extensions of (roug hly ) the wa tery stuff ar ound here falls and water are identical in the ac tual world. What makes natur al kind conce pts unique is that they are ne c e ssa ril y associate d with the two proper ties they are . I call this the specia l fea ture of natura l kind concepts. B e fo re Kr ipk e (1 98 0) a nd Put na m ( 19 75 ) d isc ov e re d th a t na tur a l ki nd c on c e pts a re ne c e ssa ri ly a sso c ia te d w ith tw o d if fe re nt k ind s o f p ro pe rt ie s, it s e e me d a s if the on ly 33 ne c e ssa ry c on ne c tio n b e tw e e n c on c e pts a nd the pr op e rt ie s th e y a re a sso c ia te d w ith is something w e ca n know only a priori. So, it looked as if possessing a conc ept re quires only a priori knowle dg e. Since na tural kind conc epts are nece ssarily associate d with two dif fe re nt k ind s o f p ro pe rt ie s, on e ma y a rg ue tha t po sse ssi on of na tur a l ki nd c on c e pts re qu ir e s k no wi ng bo th o f t he m. I f w e ha ve to k no w b oth of the m, a nd we c a n o nly c ome to k no w t he pr op e rt y in t he a c tua l w or ld t ha t a na tur a l ki nd c on c e pt p ic ks ou t a po ste ri or i, it looks as if conce pt possession may require knowing something a bout the world. Ar g ua bly , th is i s d isa str ou s f or tr a dit ion a l th e or ie s o f c on c e pts su c h a s th e on e s I discussed ea rlier, since acc ording to those theories we need not know a ny thing a bout the world to possess a c oncept. F urthermor e, most philosophers have thoug ht that they way the a c tua l w or ld i s is ir re le va nt t o th e a pp lic a tio n c on dit ion s f or ou r c on c e pts . Al tho ug h I wi ll d isc us s Pu tna m' s T wi n E a rt h th ou g ht e xpe ri me nt i n d e ta il i n C h a p t e r 3 i n t h e s e c t i o n e n t i t l e d , " P u t n a m ' s T w i n E a r t h T h o u g h t E x p e r i m e n t , " I w i l l s a y a littl e her e about how it ser ves as the f oundation of the tra ditional account of natura l kind

PAGE 64

55 I us e ' me a nin g of a te rm ' a nd ' a pp lic a tio n c on dit ion s f or a c on c e pt' 34 intercha ng eably . I a m us ing ' int e ns ion ' he re a s F re g e us e s ' Be gr iff '. 35 conce pts without t alking about the thoug ht experiment itself. The Twin Ear th thought e xpe ri me nt i s su pp os e d to sh ow tha t th e re is m or e to t he me a nin g of a te rm tha n w ha t is 34 in a spea ker' s head ( that is, the sense of a term ), g rasp of which is a matter of being in a 35 cer tain narr ow psy cholog ical state, doe s not determine its extension (that is, the refe rent of a te rm). What deter mines a term' s ext ension, at lea st in some cases, is the wa y the world ac tually is. People on Earth a nd Twin Ear th are in the same na rrow psy cholog ical state with respe ct to what they call ' water ' , y et they do not refe r to the same kind of stuff; 2 the stuff on Ea rth happe ns to be made of H O, while the stuff on Twin Ear th is made up 2 of XY Z . N o a pr ior i r e a so nin g a bo ut water wi ll g e t us to t he c la im t ha t it is H O; so me empirica l investiga tion must be done to f ind that out. 'Water ' as uttere d by Earthling s on Earth a nd ' water ' as uttere d by Twin Ear thlings on Twin Ea rth have the same intension, but differ ent extensions. One aspec t of the mea ning of natura l kind terms such as ' water ' depends on w hat things a re like in the a ctual wor ld, so one aspec t of the mea ning of what Ea rt hli ng s c a ll ' wa te r' on Ea rt h a nd Tw in E a rt hli ng s c a ll ' wa te r' on Tw in E a rt h is dif fe re nt. Many people ha ve take n from the Tw in Earth thoug ht experiment the standard a c c ou nt o f a po ste ri or i ne c e ssi tie s. At thi s p oin t it wi ll b e us e fu l to e xpla in i n mo re de ta il 2 wh a t is g oin g on wi th r e sp e c t to na tur a l ki nd te rm s su c h a s ' wa te r' . T ha t w a te r i s H O i s said to be an a posteriori nec essity ; thus, the conce pt of water is a natura l kind concept. 2 Since wate r is H O, it must be; that is, i t is necessa ry , but it is also essentially empirica l. I t is supposed by many philosophers that this kind of nece ssity (some ca ll it m etaphy sical

PAGE 65

56 I dis c us s th e tr a dit ion a l th e or y of me a nin g in m or e de ta il i n th e se c tio n in 36 Chapter 3 e ntitled, "Putnam' s Twin Earth Thoug ht Ex periment." Although this is not how Chalmers explains the relation betwe en a pr iori and a 37 posteriori nec essity , it seems to me to be a na tural wa y to understand wha t he is g etting at. See Chalmers 1996, 56-6. nece ssity ) is the same a s the kind of nec essity that all bache lors are unmarrie d is. The only differ ence is that in the bache lor ca se we can know a priori ( that is, without a pp e a lin g to a ny pa rt ic ula r e xpe ri e nc e , th a t a ba c he lor is u nma rr ie d) bu t in the wa te r c a se 2 we c a n o nly kn ow a po ste ri or i th a t w a te r i s H O. Ac c or din g to t he sta nd a rd a c c ou nt, there are interesting a priori truths we can a rrive a t about natura l kinds, for example, involving their supe rficia l properties, but ther e ar e also some ne cessa ry truths about them tha t w e c a n a rr ive a t on ly a po ste ri or i. We c a n k no w a pr ior i, f or e xamp le , th a t if the re is any water , it is t he wa tery stuff ar ound here . The discove ries about na tural kind conc epts have led to revisions of the traditional theory of mea ning, which ser ves as the ba ckg round for traditional theories of 36 conce pts. W here the traditional theory of mea ning sta tes that the intension of a te rm determine s its ex tension full-stop, the standard a ccount of natura l kind terms holds that what a pr iori nece ssary truth a g iven sentenc e expresses is dete rmined by a posterior i 2 ne c e ssa ry tr uth s g ive n th e wa y thi ng s a re in t he a c tua l w or ld. F or e xamp le , th a t H O i s 37 2 the wate ry stuff ar ound here deter mines that H O is water since wa ter jus t is the wate ry stuff ar ound here . When we say that a g iven sentenc e expresses a n a posterior i nece ssary tr uth , w ha t w e me a n is tha t it is t ru e in e ve ry po ssi ble wo rl d, bu t w e c a n o nly dis c ov e r i t e mpi ri c a lly . A c c or din g to t he sta nd a rd a c c ou nt o f n a tur a l ki nd c on c e pts , a sp e a ke r h a s to know what ter ms ex pressing natura l kind concepts ty pically apply to. I n the ca se of ' wa te r' , f or e xamp le , w e ma ke a n a pr ior i in fe re nc e tha t w a te r i s th e c le a r d ri nk a ble liq uid

PAGE 66

57 This is, of course, just an ove rsimplified example of how one mig ht give the 38 superf icial prope rties of wa ter. Th is i s n ot t o s a y tha t it is n ot c on c e iva ble , in a ve ry we a k s e ns e , th a t w a te r i s 39 2 not H O in the ac tual world. Afte r all, we could discover some day that, in fac t, we have done our e mpirical investig ation all wrong . tha t f ill s o ur la ke s a nd ri ve rs (h e re a ft e r, I wi ll r e fe r t o it a s ' the wa te ry stu ff ' ). We a lso 38 22 know a pr iori that if the wate ry stuff is H O, then wa ter is H O. We know these a priori beca use (a ) we know that wate r is the wate ry stuff, and ( b) we are justified in believing tha t w a te r i s th e wa te ry stu ff in a wa y tha t do e s n ot d e pe nd on ou r e xpe ri e nc e s. Th e se 2 conditionals and others like the m do not by themselves dete rmine that wa ter is H O. Rather, the r elation betwe en the intension and the extension of 'wa ter' is as follows: give n 22 that in the actua l world, the wa tery stuff is H O, wate r is H O in ever y possible world. F or e xamp le , if the re is w a te ry stu ff a nd it i s u nif or m, t he n th a t st uf f i s w a te r. B ut i t is 39 2 on ly a po ste ri or i th a t w e c a n d e te rm ine tha t w a te r i s H O i n th e a c tua l w or ld. I t is thi s that traditional theories of conce pts fail to acc ount for, but for w hich they should. I n th is d iss e rt a tio n, I wi ll o ff e r t he be g inn ing s o f a tr a dit ion a l a c c ou nt o f c on c e pts that acc ounts for the spec ial fea ture of na tural kind conc epts, namely , that one of the pr op e rt ie s th e y a re a sso c ia te d w ith is o ne we c a n o nly kn ow a po ste ri or i. I do thi s b y fi rs t g iving a n acc ount of conc ept individuation in the nex t chapter and later by g iving a n acc ount of conc ept possession that require s individuals to be sensitive to facts about the e xter na l w or ld ( in a wa y tha t I wi ll d e sc ri be in d e ta il) , b ut t ha t do e s n ot r e qu ir e the m to know any particula r fa cts about the e x terna l world. This acc ount of conc ept possession will work nicely with the fra mework of thought e x periments that I provide in Chapter 2 a nd Cha pte r 3 of the dis se rt a tio n. I t w ill a lso he lp p ro vid e a n a na ly sis of the zomb ie

PAGE 67

58 tho ug ht e xpe ri me nt t ha t pr ov ide s th e ph y sic a lis t w ith a n e xpla na tio n o f w hy we se e m to g e t th e a nti -p hy sic a lis t r e su lts we do fr om i t.

PAGE 68

59 CH APT ER 2 F RA ME WORK F OR DE F EN DI NG A T RA DI TI ON AL APP RO AC H I n Chapter 1, I introduced some c omplications that natural kind conc epts prese nt fo r a tr a dit ion a l th e or y of c on c e pts a nd su g g e ste d w a y s in wh ic h o ne ma y be a ble to s til l defe nd a tra ditional account of conce pts. I n this chapter, I will endorse a n acc ount of conce pts that acc ommodates a poster iori nece ssities and the specia l fea tures of na tural kind conce pts, which nonethele ss retains many important fea tures of tra ditional theories of conc epts. I n doing this, I will ex plain how I think concepts should be individuated, that is, I will ex plain what make a conc ept the par ticular c oncept it is. Once this account is in place, I will ex plain what I mean whe n I talk about conc eptual truth, conc eptual knowledg e, and c onceptua l analy sis. This will serve as par t of the fr amewor k I offe r for defe nding a traditional ac count of thoug ht experiments and conce ptual analy sis. Then, I will offer a fr amewor k for thinking about and e valuating thought e x periments, a fra mework tha t meshes with the ac count of c oncepts I have e ndorsed. On the fr amewor k I develop, thoug ht experiments can be r epre sented in terms of the experimenter, the participa nt, and three elements: a ta rg et sce nario de scribed by the experimenter; a c anonica l question about the scena rio, posted by the experimenter to the pa rt ic ipa nt; a nd the ba c kg ro un d in fo rm a tio n th a t a pa rt ic ipa nt m a y or ma y no t br ing to bear in answer ing the canonic al question. Using this framew ork we can a sk such questions as these: Who is a suitable participa nt in a thought e x periment? What must an experimenter do in desc ribing the targ et sce nario? What sorts of canonica l questions are

PAGE 69

60 For ease of expositi on, I will talk about terms when talking about natura l kind 1 terms/conce pts throughout this section. ' Meaning s of our ter ms' is equiva lent to 'how conce pts apply in other possible worlds.' When I use the ter m 'te rm' what I mean is ' term t h at ex p re s s es a c o n ce p t . ' c on fu sin g , a nd wh ic h a re we llpo se d? What kin d o f b a c kg ro un d in fo rm a tio n mi g ht b e in play , and how mig ht one be insuff iciently sensitive to its relevanc e? Once we a nswer these que stions, we can se t up some conditions for prope r per formanc e of a thought experiment and discuss how mee ting those c onditions can preve nt us from making mis ta ke s in pe rf or min g the m. Beginn ings of a Quasi -Traditional T heory of Concepts I f there are a posterior i nece ssities as I sug g est there are in Chapter 1 of the dis se rt a tio n, the n it loo ks a s if a str a ig htf or wa rd ly tr a dit ion a l th e or y of c on c e pts is pr ob le ma tic be c a us e , f or na tur a l ki nd c on c e pts , th e wa y thi ng s a re in t he a c tua l w or ld i s rele vant to their applica tion conditions. Traditional theories of c oncepts do not make room for the rele vance of thing s about the ac tual world to the applica tion of conce pts. Tr a dit ion a lly , p hil os op he rs ha ve tho ug ht t ha t th e wa y thi ng s a re in t he a c tua l w or ld i s ir re le va nt t o w he the r a c on c e pt a pp lie s, fa ils to a pp ly , o r i t is ind e te rm ina te wh e the r i t do e s, a nd tha t e ve ry thi ng tha t is ne c e ssa ry fo r t he a pp lic a tio n o f o ur c on c e pts is k no wa ble a priori. I n this section, I will sketch the beg innings of w hat I call a qua si-traditional a c c ou nt o f c on c e pts , w hic h ma ke s u se of e ve ry thi ng a tr a dit ion a l th e or y of c on c e pts makes use of , but that relies on a twodimensional semantics for terms that express our c on c e pts . T wo -d ime ns ion a l se ma nti c s sh ow s h ow the wa y thi ng s a re in t he a c tua l w or ld is r e le va nt t o th e me a nin g s o f o ur te rm s. 1 I n th is s e c tio n, I wi ll i ntr od uc e a tw odim e ns ion a l f ra me wo rk a nd e xpla in h ow it i s r e l e v a n t t o t h e k i n d o f t h e o r y o f c o n c e p t s I w a n t t o e n d o r s e . In d o i n g t h i s , I w i l l s a y a

PAGE 70

61 bit about how we should individuate conc epts. This is important, since on a twod i m en s i o n al p i ct u re , ea ch t er m h as as s o ci at ed wi t h i t t wo d i ff er en t k i n d s o f m ea n i n g, and, as a result, the conc ept expressed by the term may are associate d with two differ ent pr op e rt ie s. I t is he re wh e re the kin d o f a c c ou nt o f c on c e pts I wa nt t o e nd or se de pa rt s fr om t ra dit ion a l th e or ie s o f c on c e pts . I n the next chapter, in a section entitled, " Suitable Candidates and Conce pt Possession," I will make use of the two-dimensional fra mework I sketch he re to provide an ac count of the r ole of e mpirical informa tion in concept possession. I will not arg ue tha t on e ne e ds to h a ve inf or ma tio n a bo ut t he a c tua l w or ld t o p os se ss a c on c e pt, bu t, ra the r, tha t on e ne e ds to b e se ns iti ve to i nf or ma tio n a bo ut t he a c tua l w or ld i n a c e rt a in wa y . T he a c c ou nt o f c on c e pt p os se ssi on tha t I pr ov ide in t his dis se rt a tio n is on e tha t w ill help explain why I think we g et the wr ong result in the zombie thought experiment, and, pe rh a ps , o the rs . Intension s and Tw o-Dim ensional Sem antics Philos ophers who of fer a two-dimensional se mantics often talk a bout a term having associate d with it two differe nt intensions. W e should, then, explain what an intension is before ske tching a two-dimensional se mantic picture . One wa y of explaining what an intension is is as a f unction from wor lds to ex tensions. That is, once we specify a wo rl d in wh ic h a te rm is u tte re d, the e xten sio n o f t he te rm is d e te rm ine d. Th e mos t intuiti ve wa y of under standing what an intension is as a meaning , or, in Fr eg ean te rms, as a sense , which dete rmines re fer ence . Talk of intensions is releva nt to my projec t in that there is a way of under standing what it means for something to be a func tion from worlds to ext ensions that makes room f or a te rm (that expresses a conce pt) to have two int e ns ion s, e a c h o f w hic h d e pe nd s o n h ow a re fe re nt i s f ixed . I f t e rm s ( tha t e xpre ss

PAGE 71

62 See 2 Chalmers 1996, 56f f or a more detailed discussion of twodimensional se ma nti c s. I should point out that for some te rms the primary and sec ondary intension are 3 the same. Thoug h this is s urely oversimplified, it suffice s for my purposes he re. 4 conce pts) have two intensions, then we can a ccommodate the spec ial fea ture of na tural kin d c on c e pts . Ac c or din g to a tw odim e ns ion a l se ma nti c pic tur e , de pe nd ing on wh ic h w or ld 2 (that is, the ac tual world or some c ounterfa ctual wor ld) we a re f ix ing the ref ere nce of a term in, the term ha s a diffe rent kind of intension . A term' s primary intension has 3 nothing to do with empirica l fac ts. For e x ample, in the ac tual world it turns out that 22 water picks out H O, so the conc ept of water ref ers to H O in the ac tual world. I f, on the other ha nd, it turns out that the refe renc e of ' water ' in some world is XYZ , then ' water ' pic ks ou t X YZ in t ha t w or ld ( a t a loc a tio n a nd tim e ). We m ig ht s a y tha t w a te r i s wh a te ve r t he wa te ry stu ff is i n a wo rl d. Th e pr ima ry int e ns ion is w ha t w e or din a ri ly 4 (traditionally ) re fer to as the mea ning of a ter m, and we know a ter m's pr imary intension a priori. Ar g uably , the primary intension of ' water ' is the water y stuff ar ound here . Th e se c on da ry int e ns ion of a te rm (t ha t e xpre sse s a c on c e pt) is w ha t is re le va nt t o the spec ial fea ture of na tural kind conc epts. A sec ondary intension is the function that results from a ssuming that the r efe renc e of a term is fix ed in re lation to the actual wor ld, so that in every possible world, the re fer ence of the ter m is the same. For example, give n 22 that in the actua l world ' water ' picks out H O, ' water ' picks out H O in ever y world. So, 2 the sec ondary intension of ' water ' is ex presse d by (is a func tion from worlds to) H O. An oth e r w a y to u nd e rs ta nd a te rm ' s se c on da ry int e ns ion is a s f oll ow s: e a c h te rm is associate d with a func tion that takes it from one c ontext of uttera nce, w here a conte x t of

PAGE 72

63 I sh ou ld p oin t ou t he re tha t w he re I a m in te re ste d in se c on da ry int e ns ion s, 5 Cha lme rs , w ho a c c e pts a tw odim e ns ion a l se ma nti c pic tur e is p ri ma ri ly int e re ste d in pr ima ry int e ns ion s. utteranc e spec ifies a pa rticular w orld in which the ter m is utt ere d, to another f unction, namely , a func tion from worlds to extensions. The secondar y intension of a natur al kind term depe nds on empirical f actor s, so we ca n only know the sec ondary intension of a te rm (t ha t e xpre sse s a c on c e pt) a po ste ri or i. I t is natural for some one to under stand Kripke' s (1980) a nd Putnam's (1975) discovery about natura l kind terms in terms of a two-dimensiona l semantic picture . The thi ng we tr a dit ion a lly a sso c ia te wi th t he me a nin g of a te rm is i ts p ri ma ry int e ns ion , b ut, once w e re cog nize that there a re na tural kinds terms, we c an see that terms have two differ ent kinds of meaning s associate d with them. I f we acc ept a twodimensional picture of semantics, then we can se e how something we c an only know a poster iori can be one me aning of a te rm. I t is the nece ssary connec tion between a natura l kind term and what it picks out in the actua l world that should make us think that there is a n a posterior i e le me nt t o it s me a nin g . B e fo re Kr ipk e a nd Put na m' s d isc ov e ry a bo ut n a tur a l ki nd te rm s , philosophers thoug ht that the meaning s of our ter ms are e x hausted by what we can discover a priori. Although most philosophers foc us on the primary intension of terms when ta lking about meaning , what I am most interested in for the purposes of this diss erta tion is t he se c on da ry int e ns ion s o f n a tur a l ki nd te rm s. I a rg ue in C ha pte rs 3 a nd 4 th a t it is p la us ibl e we ha ve made mistakes in thought experiments involving na tural kind terms by not being sensitive to the a posterior i aspec t of their mea ning. What I need to do, then, is sketc h a 5 the or y of c on c e pts tha t ma ke s r oo m f or the a po ste ri or i a sp e c t of me a nin g , w hic h w ill

PAGE 73

64 require a slig ht depar ture fr om traditional theories of c oncepts. I n particula r, inc or po ra tin g te rm s' se c on da ry int e ns ion s in to a the or y of c on c e pts re qu ir e s so me re vis ion to t ra dit ion a l th e or ie s a bo ut h ow c on c e pts sh ou ld b e ind ivi du a te d. Th is i s be c a us e on a tw odim e ns ion a l se ma nti c pic tur e the sa me c on c e pt m a y , if oc c ur ri ng in on e c on te xt, be a ssi g ne d to a dif fe re nt s e c on da ry int e ns ion tha t it wo uld be a ssi g ne d if it oc c ur re d in a dif fe re nt c on te xt. I n th e ne xt se c tio n, I wi ll e xpla in h ow I thi nk c on c e pts sh ou ld b e ind ivi du a te d a nd ho w t ha t c ha ng e s tr a dit ion a l th e or ie s o f c on c e pts in s ome way s while prese rving most of the tradition. Ind ividu ating Concepts I n traditional theories of conce pts, conce pts are individuated by the prope rties with which they are associate d. I f conc epts sometimes are associate d with more than one pr op e rt y a s K ri pk e (1 98 0) a nd Put na m ( 19 75 ) a rg ue na tur a l ki nd te rm s d o, the n c on c e pts cannot be individuated that way . Consider, for e x ample, a T win Earth kind of c ase w here 2 the wate ry stuff is XYZ , and not H O. Fr om the Twin Ear th thought experiment (whic h I e xpla in i n d e ta il l a te r i n th e dis se rt a tio n) , w e le a rn tha t th e re is n o w a te r o n T wi n E a rt h 2 be c a us e wa te r i s n e c e ssa ri ly H O. B ut, it s e e ms t ha t th e re is a se ns e in w hic h th e re is water on Twin Ear th. We might say that the conc ept of wa ter, or ' water ' , picks out two 2 differ ent prope rties, the prope rty of being H O and the pr operty of being XYZ . Or, rathe r, we might say that there is watery stuffy on both Earth and T win Earth, but wha t constitutes them are distinct stuffs. Tr a dit ion a lly , p hil os op he rs ha ve us e d w ha t I c a ll a fi ne -g ra ine d a pp ro a c h to conce pt individuation, which indivi duates a conce pt acc ording to its i ntension, to which we ha ve a pr iori acc ess. On this approa ch, conc epts x and y are the same c oncept if a nd only if objects fa lling under x and y have the same prope rty . For example, the conce pt of

PAGE 74

65 a whi n and the c oncept of a fur ze a re the sa me c on c e pt s inc e the y bo th h a ve the pr op e rt y of being (roug hly ) a g ree n spiny shrub that g rows in Europe and vice versa . Alterna tively , we c an say that two conce pts are one and the sa me iff they g ener ate the sa me fu nc tio n f ro m w or lds to e xten sio ns . I wa nt t o p ro po se a c oa rs e -g ra ine d a pp ro a c h to c on c e pt i nd ivi du a tio n, a c c or din g to w hic h a c on c e pt' s se c on da ry int e ns ion is r e le va nt, tha t is , a c on c e pt i s in div idu a te d r e la tiv e to w ha t w or ld i t is us e d in . I a rg ue tha t c on c e pts do not pick out a sing le prope rty ; they only pick their prope rty out relative to a context of utteranc e (something which involves a w orld, a spea ker, a time, and a loca tion). For example, the conce pt of the wat e ry stu ff , or the use of ' water ' , which expresses the wat e ry 2 stu ff , on Earth picks out H O. The c oncept of the wat e ry stu ff , or the use of ' water ' , on Twin Ear th picks out XYZ . But, they are still the same c oncept. On this kind of view, then, x and y are the same c oncept iff x and y g ener ate the sa me func tion from context of utteranc e (in the te chnica l sense) to func tions from worlds to ext ensions. This approa ch to c on c e pt i nd ivi du a tio n le a ds na tur a lly to t he a c c ou nt o f c on c e pt p os se ssi on I pr ov ide in the ne xt ch a pte r a nd pla y s a n im po rt a nt r ole in w ha t I ha ve to s a y a bo ut t he zomb ie thought e x periment in Chapter 5. I n the next section I will clarify the notions of conce ptual truth and conc eptual knowledg e. We should g et clea r on conc eptual truth and know ledg e for at least two rea sons. First, they are centr al in discussing c onceptua l analy sis, since when w e do conce ptual analy sis, we claim to be looking for c onceptua l truths and to be ac quiring conce ptual knowledg e whe n we do it succ essfully . Second, the ter m 'c oncept' is at the r o o t o f t h e s e e x p r e s s i o n s , a n d , a s a r e s u l t , w e n e e d t o b e c l e a r w h a t t h e r e l a t i o n i s , i f a n y, betwee n conce pts and conc eptual truth and know ledg e. I n addition, we should g et clea r on the re lation betwee n conce ptual knowledg e and the world. I do not want to say that 22 wa te r i s H O i s a pie c e of c on c e ptu a l kn ow le dg e . T ha t w a te r i s H O i s a bo ut t he wo rl d

PAGE 75

66 I should point out that I do not want to say that there is no role for non6 conce ptual knowledg e, that is, empirica l knowledg e, in conc eptual ana ly sis. I n the se c tio n e nti tle d, " Sui ta ble Ca nd ida te s a nd Con c e pt P os se ssi on " of the ne xt ch a pte r, I wi ll a dd re ss t he ro le of no nc on c e ptu a l kn ow le dg e in c on c e ptu a l a na ly sis . a nd , f ur the rm or e , it is r e le va nt t o th e a pp lic a tio n o f ' wa te r' . I n a dd re ssi ng thi s is su e , it is g oing to be important to ge t clear on how it is that conceptua l knowledg e is g rounded. Co nc e pt ual Tr ut h, C onc e pt ual K nowle dge , a nd C onc e pt ual Ana lys is Con c e ptu a l tr uth s a re wh a t w e a re loo kin g fo r w he n d oin g c on c e ptu a l a na ly sis , a n d i n d o i n g c o n c e p t u a l a n a l ys i s , w e h o p e t o g a i n c o n c e p t u a l k n o w l e d g e , t h a t i s t o s a y, knowledg e of c onceptua l truths. Only afte r we deter mine what conc eptual truths and 6 conce ptual knowledg e ar e ca n we de termine whe ther ther e ac tually are conce ptual truths a nd wh e the r w e ha ve or c a n h a ve a ny kn ow le dg e of the m. M y g oa l in thi s se c tio n is to explain what conce ptual truth, conce ptual knowledg e, and c onceptua l analy sis are, a nd the re lation among them. This will help us to understand what we are try ing to do whe n we a ppeal to thoug ht experiments, and make c lear what kinds of truths count as conce ptual ones. I t will also help us to understand the conne ction betwee n a priori knowledg e and ne cessity , and wha t kind of knowledg e the knowle dg e of a posteriori nece ssities amounts to or does not amount to. This i s important to the rest of the dissertation, since I think that what is a posteriori nec essar y is releva nt to the outcome of some thoug ht experiments. Conceptual Truth an d Conceptual Know ledge Most (if not all) philosophers who fa vor conc eptual ana ly sis ag ree that the kind of knowledg e we acquir e by doing c onceptua l analy sis is conceptual knowle dg e. Howe ver, it i s n ot a lw a y s c le a r w ha t th e y me a n b y ' c on c e ptu a l kn ow le dg e ' oth e r t ha n th a t it is kn ow le dg e we a rr ive a t in do ing c on c e ptu a l a na ly sis , o r, pe rh a ps , k no wl e dg e tha t is g ro un de d in the sa me wa y tha t kn ow le dg e we a rr ive a t w he n w e do c on c e ptu a l a na ly sis is

PAGE 76

67 I want to make r oom for conc eptual truths and c onceptua l knowledg e that we do 7 not arrive at by doing c onceptua l analy sis. See my examples of the color r ed, below. I say that conce ptual knowledg e is conc eptually g rounded, but one might think 8 there is an ambig uity here about what is ac tually g rounded—the conce ptual truth itself or our knowledg e of it. Afte r all, conc eptual knowledg e is knowledg e of a conce ptual truth. I want to say that our knowledg e of the truth (not the truth itself) is g rounded conce ptually (that is, a priori) , since wha t I am re ally conce rned w ith in thi s dissertation is w ha t ki nd s o f t hin g s w e c ome to k no w w he n w e do c on c e ptu a l a na ly sis . Th e re is a c on ne c tio n b e tw e e n a pr ior i kn ow le dg e a nd ne c e ssi ty tha t I sh ou ld 9 briefly mention here that many philosophers ac cepte d befor e Kr ipke and Putnam dis c ov e re d a po ste ri or i ne c e ssi tie s, na me ly , th a t th os e a nd on ly tho se thi ng s th a t w e c ome to know a priori a re ne cessa ry . That cha ng ed, howe ver, w hen Kr ipke and Putnam 2 showed that wa ter is nec essar ily H O, and we can only come to know it a poster iori. So, g rounded. There seem to be a t least two way s of cha rac terizing c onceptua l knowledg e: 7 e ith e r b y its su bje c t ma tte r o r t he wa y in w hic h it is k no wn . I n th is s e c tio n, I wi ll e xpla in how I use the e x pression ' conce ptual knowledg e' and why conce ptual knowledg e is not kn ow le dg e a bo ut c on c e pts pe r se , b ut, ra the r, kn ow le dg e a bo ut t he wo rl d th a t is g ro un de d in a pa rt ic ula r w a y , n a me ly , a pr ior i. 8 We can disting uish two way s of picking out the rele vant kind of g rounding of knowledg e: 11 (C K ) x ha s c on c e ptu a l kn ow le dg e tha t p i ff p is a c on c e ptu a l tr uth a nd x kno ws that p. I stated above that what is most important about conce ptual knowledg e is the wa y that 1 the knowledg e is g rounded, not the c ontent of the knowle dg e cla im. (CK ) lea ves it open 1 for x to have lear ned that p either a priori or a posterior i. That is, (CK ) is neutra l on 1 what g rounds x's knowledg e that p. Obviously (CK ) does not ca pture the ide a that one of the thing s that is special about c onceptua l knowledg e is how it is gr ounded. Another no tio n o f c on c e ptu a l kn ow le dg e tr e a ts i t w ho lly a s a ma tte r o f h ow wh a t is kn ow n is known. 22 (C K ) x ha s c on c e ptu a l kn ow le dg e tha t p i ff x kno ws tha t p a pr ior i. 9

PAGE 77

68 althoug h it looks as if all a priori knowledg e is of thing s that are nece ssary , there are t h i n g s t h a t w e c a n o n l y c o m e t o k n o w a p o s t e r i o r i t h a t a r e e q u a l l y n e c e s s a r y. I should point out here a diffe renc e betwe en conc eptual and nonconce ptual 10 knowledg e. Nonconce ptual knowledg e is knowledg e that we do not come to a pr iori. We must acquire nonconce ptual knowledg e by rely ing on pa rticular e x perie nces tha t we have. A n example of something tha t I know non-c onceptua lly is that I am hung ry now. We can ha ve non-c onceptua l knowledg e of pr opositions t hat we c an, in principle, know a priori. That is, there are some propositions we ca n know both conce ptually and nonconce ptually . On thi s a c c ou nt, a ll a pr ior i kn ow le dg e is c on c e ptu a l kn ow le dg e a nd vic e ve rs a . I sh ou ld ma ke c le a r h e re wh a t I me a n w he n I sa y tha t so me on e kn ow s so me thi ng a pr ior i: (A Pr ior i K no wl e dg e ) x kn ow s th a t p a pr ior i if f ( a ) x kn ow s th a t p a nd (b ) x is j u s t i fi ed i n b el i ev i n g t h at p i n a w ay t h at d o es n o t d ep en d ep i s t em i ca l l y o n x 's e xpe ri e nc e s. W h en I s ay t h at x i s j u s t i fi ed i n b el i ev i n g t h at p i n a w ay t h at d o es n o t d ep en d o n x 's experience s, I do not mean to imply that x need not ha ve any experience whatsoeve r (f or that would not make any sense) . We need some e x perie nce to g ain conc epts; but once we ha ve the c oncepts, no e x perie nce a t all is required f or conc eptual knowledg e. No particula r experienc e would help x in jus tify ing his be lief that p. F or example, knowing tha t r e d is a c olo r i s n ot b a se d o n a ny pa rt ic ula r e xpe ri e nc e , b ut i t se e ms t ha t w e ha ve to have some visual experience to know that red is a c olor. What seems to make it a prior i kn ow le dg e is t ha t it is n ot w a rr a nte d b y the c on te nt o f a ny e xpe ri e nc e . A c c or din g to 2 (C K ), on e do e s n ot h a ve c on c e ptu a l kn ow le dg e of a pr op os iti on un le ss o ne c a me to kn ow it a pr ior i. I wi ll g ive e xamp le s b e low of ho w i t is po ssi ble to k no w s ome propositions either a pr iori or a poster iori. Ex amples like the one s I will provide leave ro om f or the po ssi bil ity of on e pe rs on kn ow ing tha t p a pr ior i, a nd a no the r k no wi ng it a p o s t e r i o r i , t h a t i s , c o n c e p t u a l l y a n d n o n c o n c e p t u a l l y, r e s p e c t i v e l y. 10 2 Conceptual knowle dg e is chara cter ized wholly epistemically . As a re sult, one mig ht a rg ue tha t it is u nc le a r w ha t is c on c e ptu a l a bo ut i t. T ra dit ion a lly (i n a na ly tic

PAGE 78

69 philosophy to the midpoint of the twentieth century ) thing s we c an know a priori have been tr eate d as conc eptual knowledg e. F urther, it is not an ac cident that the tra dition has trea ted a pr iori knowledg e as c onceptua l knowledg e. F or it has bee n thought, I believe corr ectly , that the explanation of the possibilit y of a pr iori knowledg e depe nds on that kn ow le dg e be ing g ro un de d in a c e rt a in w a y in o ur c on c e pts . ( Of c ou rs e , K a nt f a mou sly denied that a ll a priori knowledg e wa s conce ptual. But by and lar g e the tra dition has not sided with Kant on this matter.) The re is anothe r cha rac terization of conce ptual 3 knowledg e, name ly , that spelled out in (CK ), 33 (CK ) x has conceptua l knowledg e that p iff x knows that p on the basis of her g ra sp of the c on c e pts tha t pr op os iti on p in vo lve s, wh ic h h a s b e e n th ou g ht t o c ha ra c te ri ze a sp e c ie s o f k no wl e dg e , w hic h, be c a us e of ho w i t is gr ounded, is a prior i knowledg e. I n the abse nce of any other wa y of under standing the po ssi bil ity of a pr ior i kn ow le dg e , th e n, it w ou ld l oo k a s if a ll a pr ior i kn ow le dg e wo uld 3 be conc eptual knowledg e , or to put it in t erms of our define d notions, all conceptua l 23 2 k n o wl ed ge i s co n ce p t u al k n o wl ed ge , an d t h e l i ce n s e f o r c al l i n g co n ce p t u al k n o wl ed ge wh a t w e c a ll i t is tha t it c oin c ide s w ith a nd its po ssi bil ity is e xpla ine d b y the po ssi bil ity 3 o f c o n ce p t u al k n o wl ed ge . I n the re mainder of this section, I will mean by ' conce ptual knowledg e' , without 2 any subscripts, conce ptual knowledg e . I choose this as the ba sic notion for discussion beca use it coincides with wha t the tradition larg ely has identified a s conce ptual kn ow le dg e , b ut l e a ve s o pe n q ue sti on s a bo ut w ha t e xac tly it c ome s to . I t ma y e me rg e a s a substantive point that conceptua l knowledg e is g rounded in g rasp of conce pts, that is, that 23 conce ptual knoweldg e is conce ptual knowledg e . At this point , howeve r, we can move on to s e e wh y c on c e ptu a l kn ow le dg e is n ot k no wl e dg e a bo ut c on c e pts .

PAGE 79

70 Once we de termine wha t it would mean for c onceptua l knowledg e to be a bout c on c e pts , w e wi ll s e e tha t c on c e ptu a l kn ow le dg e is n ot a bo ut c on c e pts pe r se . I wi ll explain our ordinary understanding of conc eptual knowledg e and show why it does not make se nse to say that conce ptual knowledg e is knowledg e about c oncepts by introducing examples of conce ptual analy sis, ex amples of c onceptua l knowledg e, and examples of non-conc eptual knowledg e. The se examples will ill ustrate that the sta tus of k n o wl ed ge a s co n ce p t u al k n o wl ed ge d o es n o t d ep en d o n t h e c o n t en t o f t h e k n o wl ed ge claim, but on how we c ome to know it. I f conc eptual knowledg e we re a bout conce pts, then it seems that all conce ptual kn ow le dg e wo uld c ome to i s k no wi ng a c or re c t a c c ou nt o f v a ri ou s so rt s o f r e le va nt f a c ts a bo ut c on c e pts (t ha t is , e ith e r k no wi ng a c or re c t th e or y of c on c e pts or kn ow ing c e rt a in fac ts about particular conce pts). But, we think we have at least some c onceptua l knowledg e, and, g ener ally , we do not think that we ha ve ar rived a t the corr ect theor y of co n ce p t s . I n ad d i t i o n , wh en we t h i n k o f c o n ce p t u al k n o wl ed ge a s t h e k i n d o f k n o wl ed ge we a cquire when we do philosophy , it seems that we think of knowle dg e that is much broade r in scope tha n just knowledg e about c oncepts. F or example, many think that the knowledg e we have of many mathematica l truths is conceptual knowle dg e, but we do not think that this knowledg e is about conc epts. I t seems that I know a pr iori, for e x ample, tha t 1 i s g re a te r t ha n 0 . H ow e ve r t his is g ro un de d, it d oe s n ot s e e m to be g ro un de d in experience . I t is about the number 1 and the number 0, a nd about a r elation betwe en nu mbe rs , th a t of be ing g re a te r t ha n. I t is no t a bo ut c on c e pts a t a ll. So, if thi s is conce ptual knowledg e, as it would be tra ditionally classified a nd counts by my definition 2 of conc eptual knowledg e , it looks as if conce ptual knowledg e is not knowledg e about c on c e pts pe r se .

PAGE 80

71 I say ' nece ssarily ' here beca use, as I stated above , there are cer tain truths that 11 o n e p e r s o n c a n k n o w c o n c e p t u a l l y a n d a n o t h e r p e r s o n c a n k n o w n o n c o n c e p t u a l l y. This is not t o say , howeve r, that conc eptual knowledg e is neve r about c oncepts. Conceptual knowle dg e is about our c oncepts, for example, when we are doing c onceptua l analy sis on the conce pt of a conce pt , which is what the philosopher s I discussed ea rlier are doing. H oweve r, wha t is most important about the kind of knowledg e that we may ac q u i re wh en gi v i n g a p h i l o s o p h i ca l t h eo ry o f c o n ce p t s i s n o t t h e f ac t t h at t h e k n o wl ed ge is about conce pts. Rather, wha t is int ere sting is that it is a kind of knowledg e that we can acquir e without doing a ny empirica l resea rch, or without appealing to any particula r e xpe ri e nc e . T his is t he fe a tur e of c on c e ptu a l kn ow le dg e tha t se e ms t o ma ke it distinctive. Howeve r, it remains to be se en how we can g round any knowledg e that we do not nece ssarily acquir e throug h empirica l methods. 11 So, a lth ou g h w e a re a na ly zing c on c e pts wh e n w e a re do ing c on c e ptu a l a na ly sis , t h e k i n d o f k n o wl ed ge t h at we ac q u i re b y d o i n g co n ce p t u al an al y s i s i s n o t k n o wl ed ge a bo ut c on c e pts pe r se . Rather, we have to use talk about conc epts when w e talk about our wa rra nt for making these pa rticular c laims, since we w ant to say that they are conce ptually g rounded. Now that we have se en that conc eptual knowledg e is not alway s knowledg e about conce pts, we ca n see w hat it means to say that conce ptual knowledg e is knowledg e about the world that c an be a rrived a t a priori. F or example, consider the conc eptual truth, ' B a c he lor s a re un ma rr ie d ma le s' . T he se nte nc e is t ru e , a nd the re is n o r e a so n to sa y tha t it is not about world any less than ' Ba chelor s are frivolous' is about the world. The predic ates a pply to things in the world, so it looks as if someone w ould have to g ive an

PAGE 81

72 a rg ume nt t o s ho w t ha t c on c e ptu a l tr uth s a re no t a bo ut t he wo rl d. Con c e ptu a l tr uth s ju st seem to be g rounded diff ere ntly than other sor ts of truths, which is an epistemolog ical fa c t, n ot a fa c t a bo ut t he ir su bje c t ma tte r. Whe n w e do c on c e ptu a l a na ly sis , w e g e ne ra lly do not want to find out about conc epts; rather , we wa nt to find out about featur es of the world. We want to be a ble to make se nse of the w orld by cate g orizing wha t is in i t (i.e., w e a r e lo o k in g f o r s im il a r it ie s a n d d if f e r e n c e s a mo n g d if f e r e n t t h in g s in th e w o r ld ) . We do thi s b y loo kin g a t f e a tur e s o f o bje c ts a nd de te rm ini ng wh ic h o f t ho se fe a tur e s se e m to be nec essar y for that objec t to remain the objec t that it is . Whether a f eatur e is nec essar y for some obje ct is usually not something that we can a rrive a t by appea ling to any particula r experienc e. I wa nt t o d ra w a tte nti on to a pa rt ic ula r k ind of c on c e ptu a l kn ow le dg e , w hic h I c a ll conditional conce ptual knowledg e. Conditional conce ptual knowledg e is a spec ial knowledg e of the application conditions of a c oncept, a nd it comes in the form of c on dit ion a ls. Th a t is , s ome c on c e ptu a lly tr ue pr op os iti on s in vo lvi ng a c on c e pt m a y be in the form of a conditional F or example, someone may know that if the c ontext of utt e ra nc e is s uc ha nd -s uc h, the n th e a pp lic a tio n c on dit ion s f or the c on c e pt i n q ue sti on is su c ha nd -s uc h. Th is w ill be c ome imp or ta nt i n my dis c us sio n o f P utn a m' s T wi n E a rt h thought e x periment, whe re I show this kind of conditional conce ptual knowledg e via a n example, and in my analy sis of the zombi e thoug ht experiment. Conceptual Analysis and Con ceptual K now ledge The wa y in which philosophers primar ily acquir e and ve rify conce ptual knowledg e is by doing c onceptua l analy sis. At this point, we should g et clea r on the traditional view of c onceptua l analy sis and what a c onceptua l analy sis may look like. There is also conce ptual knowledg e that we need not a cquire by doing c onceptua l

PAGE 82

73 a na ly sis ; th a t is , w e c a n h a ve c on c e ptu a l kn ow le dg e e ve n if we do no t ha ve a fu ll analy sis of a g iven conc ept. Fur thermore , as I stated above , there is some knowledg e we ma y a c qu ir e e ith e r b y do ing a na ly se s o r b y do ing e mpi ri c a l in ve sti g a tio n, a nd , a s a re su lt that knowledg e is both conce ptual and non-c onceptua l knowledg e. Sometimes when I talk about conc eptual ana ly sis, I am talking about the statement we g et as a result of a complete c onceptua l analy sis, which is of the following fo rm : ~ ( œ x )(x is P iff . . .). R Ca ll t his c on c e ptu a l a na ly sis . A c on c e ptu a l tr uth , o r a bit of c on c e ptu a l kn ow le dg e wi ll appea r on the rig ht hand side of the bic onditional. There a re two w ay s that I may have conce ptual knowledg e of a conce ptual analy sis: either I know what the c onceptua l RR analy sis expresses and why it is t rue, or I may know that it is a conce ptual analy sis , but R wi tho ut k no wi ng wh y it i s tr ue . T ha t is , I ma y kn ow tha t it is a c on c e ptu a l a na ly sis only on authority . Other times whe n I talk about conc eptual ana ly sis I ref er to the a c tiv ity of do ing c on c e ptu a l a na ly sis , w hic h I ta ke a s d oin g tho ug ht e xpe ri me nts . Su re ly the re a re oth e r w a y s o f d oin g c on c e ptu a l a na ly sis , b ut I a m no t in te re ste d in tho se in t his dissertation. Call the activity of doing conce ptual analy sis, whether tha t is doing thoug ht A experiments or something e lse, conc eptual ana ly sis . This disti nction should clear up any confusion that may have c ome up later in the dissertation. L et us consider the following canonic al forms in which w e would state w hat the a na ly sis of a c e rt a in c on c e pt i s: ~ ( œ x )( M x iff . . .) ~ ( œ x ,y )( M x y iff . . .) ~ ( œ x ,y ,z)( M x y z iff . . .) . . .

PAGE 83

74 We are restric ting our attention here to conce pts ex presse d by predic ates of one or more place s, which may or may not be complex predica tes. Clearly , these for ms in which we ex p re s s an al y s es n ee d n o t co m m i t u s t o t al k i n g about conce pts. There need be no mention in various instantiations of these forms of a ny conce pts; they are in the form of universal g ener alizations and contain predic ates that e x press c oncepts. R Th e re is a wa y tha t w e c a n r e wr ite the a bo ve fo rm s o f c on c e ptu a l a na ly sis , h o we v er , s o t h at t h e c o n ce p t u al k n o wl ed ge w e a rr i v e a t i s i n t h e f o rm o f k n o wl ed ge a bo ut c on c e pts : ~ ( œ x )(The conce pt C applies to x iff. . .), e tc. Although this is written so that it contains talk about conc epts, it is jus t another w ay of R sta tin g c on c e ptu a l a na ly se s tha t c a n b e wr itt e n d ow n w ith ou t a ny e xplic it s ta te me nts about conc epts; it is not nece ssary for us to use c oncept talk in our f ormulations of R conce ptual analy ses . What makes these c onceptua l analy ses is not that they are about conc epts but that they are biconditionals that meet the conditions require d for them to express conc eptual R analy ses . The fa ct that conc epts are expressed by terms in these sente nces doe s not, of course , mean that they are about those conc epts. We do not think t hat when w e say s o m et h i n g i n E n gl i s h t h at we ar e s ay i n g an y t h i n g about Eng lish. S imilarly , when we say s o m et h i n g u s i n g co n ce p t s , we ar e n o t n ec es s ar i l y t al k i n g about conce pts. Now let us consider an example from e pistemology , the ana ly sis of the conc ept of knowledge : ( œ x )( œ p) (x kn ow s th a t p i ff x be lie ve s th a t p, it i s tr ue tha t p, x' s b e lie f t ha t p i s justified, and x's justification for his belief that p is undefea ted).

PAGE 84

75 We will take this as a sche matic ana ly sis in li g ht of Gettier c ountere x amples, which a re illust rations of how the traditional ana ly sis of the conc ept of knowledge as a justified, true be lief fa ils. The following passag e fr om Gettier pre sents the counter example: Suppose that Smi th and J ones have applied for a ce rtain job. And suppose that Smit h has strong evidenc e for the following conjunctive pr oposition [ two propositions connecte d by an ' and' ]: (d) Jones is the man who will ge t the job, and J ones has ten c oins in his pocket. Smi th's e vidence for ( d) mig ht be that the president of the company assure d him that J ones would in the end be selec ted, and that he, Smith, had counted the coins in J ones' s pocket ten minutes ag o. Pr op os iti on (d ) e nta ils : ( e ) T he ma n w ho wi ll g e t th e job ha s te n c oin s in his po c ke t. L e t us su pp os e tha t Smi th s e e s th e e nta ilm e nt f ro m ( d) to ( e ) a nd a c c e pts (e ) o n th e g ro un ds of (d ), fo r w hic h h e ha s st ro ng e vid e nc e . I n th is c a se Smit h is clea rly justified in believing tha t (e) is true . But imag ine, furthe r, that unknown to Smith, he himself, not J ones, will g et the job. And also, unknown to Smit h, he himself has te n coins in his pocket. Proposition (e) is then true, thoug h proposition (d), from whic h Smit h inferr ed (e ), is f a lse . I n o ur e xamp le , th e n, a ll o f t he fo llo wi ng a re tr ue : ( i) (e ) i s tr ue , ( ii) Smit h b e lie ve s th a t ( e ) i s tr ue , a nd (i ii) Smit h is jus tif ie d in be lie vin g tha t ( e ) i s tr ue . B ut i t is e qu a lly c le a r t ha t Smi th d oe s n ot know that (e) is true; for ( e) is true in virtue of the number of coins Smith has in his pocket, while Smit h does not know how many coins are in Smi th's poc ket, and ba ses his belief in (e ) on a c ount of the c oin s in Jone s' s p oc ke t, w ho m he fa lse ly be lie ve s to be the ma n w ho wi ll g et the job. (Ge ttier 1963, 121-22) This countere x ample is supposed to show that someone c an have a justified, true be lief tha t is no t kn ow le dg e , a nd , th us , th a t th e tr a dit ion a l a na ly sis of kn ow le dg e , a lth ou g h it may g ive nec essar y conditions for knowledg e, does not g ive suffic ient conditions for kn ow le dg e . A lth ou g h th e no tio n o f b e ing un de fe a te d n e e ds so me c a re fu l sp e lli ng ou t in lig ht o f G e tti e r c ou nte re xamp le s, we c a n s e e ho w t he a bo ve c on c e ptu a l a na ly sis is g rounded. Clearly we would not attribute know ledg e of a proposition to som eone w ho does no t be lie ve the c on te nt o f t he pr op os iti on . N or wo uld we a c c e pt t ha t so me on e kn ow s a pr op os iti on tha t is fa lse . We c a n ma ke no se ns e of kn ow le dg e of a pr op os iti on tha t is fa lse . F ur the rm or e , w e do no t sa y tha t so me on e kn ow s th a t p i f s he on ly be lie ve s th a t p

PAGE 85

76 Th is i s o bv iou sly a sim ple e xamp le of c on c e ptu a l a na ly sis . I do no t me a n to 12 rule out fa r more c omplicated conc eptual ana ly ses. Surely we do not spend a ll of our tim e in p hil os op hy do ing the se kin ds of c on c e ptu a l a na ly se s. Of te n a c on c e ptu a l a na ly sis is only part of a n arg ument for a particula r position, and sometimes there ma y be no conce ptual analy sis in an arg ument for a position. W e may raise que stions about whether the la tte r k ind s o f a rg ume nts a re ph ilo so ph ic a l a rg ume nts a t a ll. on a hunch or on authority . Finally , in light of the G ettier e x amples, we would not say that someone knows that p e ven if the pe rson believe s that p, it is t rue tha t p, and the person is justified in believing that p, if it turns out that there is something de fec tive a bo ut t he jus tif ic a tio n— in t he c a se a bo ve , a fa lse pr e mis e , it se lf jus tif ie d, wh ic h is cruc ial in an infer ence leading to the belief in the true proposition just ified in the lig ht of t h e b e l i e v e r ' s e v i d e n c e . In t h i s c a s e t h e j u s t i f i c a t i o n i s d e f e c t i v e , o r , w e w i l l s a y, defe ated. The re is a que stion whether the only way in which a justification ca n be de fe a te d is by its inv olv ing a fa lse jus tif ie d p re mis e . So , in the a bo ve a na ly sis , ' undefe ated' is itself a term that stands in nee d of fur ther a naly sis. I say that we know the se thing s about knowledg e, and w e will all ag ree that we do, but how exactly ? Prima fac ie, it seems that we r ecog nize that knowledg e re quires cer tain things, a nd we r ecog nize this without doing any rese arc h into the conditions that ob ta in w he n p e op le ha ve kn ow le dg e . We se e m to kn ow the m a pr ior i. The only way to arrive at an a naly sis like the one above is through c onceptua l A analy sis . The e x ample of the conce pt of knowledge prese nts us with conceptual knowledg e—ther e is no way to determine w hat counts as knowle dg e by appea ling to a particula r experienc e. That is to say , althoug h I need to ha ve some kind of e x perie nce 12 A to eng ag e in conc eptual ana ly sis at all, there is no particular experience that I can ha ve R that will either tell me wha t a conc eptual ana ly sis of knowledg e is or whe ther a n R analy sis that I have c ome up with is corre ct. Howe ver, I may enter tain various thoug ht

PAGE 86

77 experiments that involve the conce pt of knowledge , f or e xamp le , th ou g ht e xpe ri me nts R like Gettier' s, to show that a ce rtain ana ly sis that I have g iven is incorre ct. Or, I may enter tain a thoug ht experiment to show that an analy sis is correc t despite a sce nario that one mig ht think would defea t an ana ly sis if any scena rio would. I will discuss the role of A th ou ght ex peri men ts in con cept ual anal y si s in th e sec ti on in th is chap ter ent it led , "A Fr amewor k for Thinking About Thoug ht Ex periments." RA No t a ll c on c e ptu a l kn ow le dg e is o f c on c e ptu a l a na ly se s . Co nc e ptu a l a na ly sis is a p ro ce s s . I t ca n re s u l t i n k n o wl ed ge t h at i s a p ri o ri b u t t h at fa l l s s h o rt o f p ro v i d i n g a R conce ptual analy sis . For example, the knowledg e that x knows that p only if it is true R that p is a priori, but it does not constitute knowledg e of a conce ptual analy sis . I t states only a nec essar y condition on someone' s knowing something. I t might help to clar ify matters and a void confusions if we introduc e two diffe rent AA no tio ns of c on c e ptu a l a na ly sis , o ne str on g e r, on e we a ke r. Str on g c on c e ptu a l a na ly sis A y ie lds ne c e ssa ry a nd su ff ic ie nt a pp lic a tio n c on dit ion s. Wea k c on c e ptu a l a na ly sis on ly trac es conne ctions betwee n conce pts (i.e., it trace s the rela tions between diff ere nt conce pts). For example, the conce pt of being triangular and the c oncept of being trilat eral a re c on c e ptu a lly c on ne c te d in the se ns e tha t bo th h a ve so me thi ng to d o w ith having three of something and in the sense that they apply to the same thing s in the A wo rl d. Ma ny ph ilo so ph e rs ha ve a rg ue d th a t th e pr a c tic e of c on c e ptu a l a na ly sis is R bankrupt be cause we ha ve fa iled to arrive at conc eptual ana ly ses for most interesting conce pts, or even f or any interesting conce pts. This seems to be the re sult of them AA understanding conce ptual analy sis as strong conce ptual analy sis . But sure ly we ha ve R had succ essful conc eptual ana ly ses of the we aker sort, as in the ca se of the disc overy of

PAGE 87

78 nece ssary conditions, at least, for know ledg e, and the ref utation of the traditional analy sis, which itself results in a bit of conc eptual knowledg e. There is another distinction to be made he re, na mely , one betwe en those a naly ses that g ive conditions in noncircular terms and those tha t do not. Both strong and we ak a na ly se s c a n b e g ive n in c ir c ula r o r n on c ir c ula r t e rm s. A s tr on g a na ly sis g ive n in no nc ir c ula r t e rm s is the fu lle st a na ly sis tha t c a n b e g ive n, bu t it is t he mos t di ff ic ult to g et. A strong analy sis in circular te rms does not seem to g ive us much of a n an aly sis at all; same g oes for a we ak ana ly sis in circular te rms. The re sults that I sug g est we a im for a re we a k a na ly se s g ive n in no nc ir c ula r t e rm s. Consider the following examples of conce ptual knowledg e that we do not acquire A by do ing str on g c on c e ptu a l a na ly sis : (1) Red is a c olor. Of c ou rs e , in thi s c a se we mus t pu t is su e s a bo ut w he the r ' is r e d' e xpre sse s a c on c e pt a t a ll aside (e .g ., worrie s that might ar ise from re cog nition that 're d' is vag ue). I f we do not want to allow ' is red' to express a conc ept, as I stated ea rlier in the disser tation, we may still talk about g iving a conce ptual analy sis of ' is red' , since c onceptua l analy sis is an idealized prac tice. Howe ver, w e should be in a position to answer the following qu e sti on : c a n s e nte nc e s e mpl oy ing pr e dic a te s th a t do no t e xpre ss c on c e pts e xpre ss c on c e ptu a l tr uth s? I n o the r w or ds , d o a ll p re dic a te s in a se nte nc e ha ve to e xpre ss conce pts for that sentenc e to express a c onceptua l truth? Since I define d a conc eptual truth as one that we come to know a priori, the answ er to the que stion is "no." What makes a conce ptual truth conce ptual i s t h e w a y w e c o m e t o k n o w i t . C o n s e q u e n t l y, conce ptual knowledg e does not re quire a ny mention of conc epts, or pre dicates tha t e xpre ss c on c e pts . Ra the r, c on c e ptu a l kn ow le dg e re qu ir e s k no wi ng so me thi ng tha t is

PAGE 88

79 g ro un de d in a pa rt ic ula r w a y . T his se e ms t o ma ke so me ro om f or se nte nc e s th a t c on ta in predic ates that do not express conc epts to express conce ptual truths. L et us consider , by wa y o f i l l u s t ra t i o n , a d i s p o s i t i o n al an al y s i s o f 'r ed '. F or a ny x, x is r e d if f x ha s th a t c olo r w hic h h a s th e dis po sit ion to c a us e in c e rt a in ob se rv e rs re d p e rc e pti on s. I do not want to endorse such an a naly sis, but I want to illustrate that we do not ne ed a strong analy sis given in nonc ircular terms of the c oncept of red to know (1). ( 1) g ives us nece ssary and suff icient conditions, but since it is circular , it is not an ana ly sis. Howeve r, (1) can be conce ptual knowledg e, and thus e x presse s a conc eptual truth, eve n A if we do not lea rn it on the basis of conc eptual ana ly sis . We can g et a we ak ana ly sis and still establish conce ptual truths. This brings us ba ck to what I said ea rlier a bout there A being proce sses of conc eptual ana ly sis that are strong er a nd wea ker. A wea k conce ptual A analy sis may g ive us something like ( 1) af ter we do a dispositional analy sis of the conce pt of red ; we nee d not ge t a strong analy sis of the conc ept of red to ge t (1), and ( 1) is not a strong noncirc ular a naly sis of red . We do, however , g et (1) a priori. This makes (1 ) c on c e ptu a l kn ow le dg e . E ve n th ou g h ' is r e d' is a va g ue pr e dic a te , w e c a n c ome to know things a bout red a priori. B ut, what we know a priori a bout it has to be in circular terms. Also consider the following example of conc eptual knowledg e that we do not a c qu ir e by do ing c on c e ptu a l a na ly sis : (2) I f this is red, then it is colored. A We cer tainly cannot g et (2) f rom conc eptual ana ly sis . Howeve r, we can know a priori that the conditional is true. I talk in more deta il about conditional conce ptual knowledg e, that is, conc eptual knowledg e that is conditional in form, later in the

PAGE 89

80 dis se rt a tio n. Th e no tio n o f c on dit ion a l c on c e ptu a l kn ow le dg e is i mpo rt a nt f or my analy ses of a varia tion of the Twin Ear th thought experiment and the zombi e thoug ht experiment. I n case s of conditionals that we know a priori, the a ntece dent of the conditional will contain something tha t we ca n only know a poster iori, but the conditional as a w hole will be something that we c an come to know a priori. Tha t the antec edent c ontains something we can only know a poster iori is not relevant to whe ther we c a n k no w t he wh ole c on dit ion a l a pr ior i. So, it l ooks as if we can ha ve conc eptual knowledg e that is not an ana ly sis of a particula r conc ept, and we may be able to have knowle dg e of c onceptua l truths ex presse d by se nte nc e s n ot e nti re ly ma de up of pr e dic a te s, wh ic h e xpre ss c on c e pts (s inc e a ll conce ptual knowledg e is is a priori knowledg e). We c an also ha ve conditional conce ptual knowledg e. Another example of conc eptual truth that we cannot a cquire A throug h conce ptual analy sis , and which looks diff ere nt from the ca nonical for mulation R of conc eptual ana ly sis I provided a bove, but which is a pr iori, is the proposition that 2 + 2 = 4. Nothing in this refer s to conce pts at all, but we ca n cer tainly arr ive at this a priori. I n addition, doing thoug ht experiments would not help us to arrive a t the proposition, and, fur ther, the pr oposition i s not an an aly sis at all. I t does not fit into the canonica l forms of a n analy sis that I g ave a bove. Fur thermore , there are examples in ge ometry of conc eptual knowledg e that we do A not acquire throug h conce ptual analy sis . For example, there a re some ne cessa ry b i co n d i t i o n al s t h at co n n ec t ci rc l es wi t h o t h er t h i n gs t h at wo u l d b e c o n ce p t u al k n o wl ed ge A that we do not g et from doing conce ptual analy sis . For example, consider the f ollowing analy sis of the conc ept of a c irc le :

PAGE 90

81 Of c ou rs e , w e no w k no w t ha t ' B ' ref ers to 3.141592... 13 ( œ x)( x is a c ir c le if f x is a se t of po int s in a pla ne e qu idi sta nt f ro m a g ive n p oin t in the plane.) I f one w ere to do a strong analy sis of the conc ept of a c irc le , o ne mig ht c ome up wi th something tha t looks li ke the de finition above, which is in the for m of a biconditional. Howeve r, there are nece ssary biconditionals about circle s that are not the result of a strong analy sis of the conc ept of a c irc le . C o n s i d er , fo r e x am p l e, t h e f o l l o wi n g: ( œ x )(x is a circle if f x's circ umfere nce = B d) R Ev e n th ou g h w e do no t g e t th is a s a c on c e ptu a l a na ly sis , w e c a n c ome to k no w t his a priori bec ause ' B ' wa s d e sig na te d to re fe r t o th e ra tio of the c ir c umf e re nc e a c ir c le to i ts dia me te r, wh a te ve r i t is . Th e dia me te r o f a c ir c le mul tip lie d b y the ra tio of its 13 circ umfere nce to the dia meter w ill get us to the distance around the circ le. Ther e is no pa rt ic ula r e xpe ri e nc e we c a n a pp e a l to a s e vid e nc e fo r d isc ov e ri ng thi s. Cle a rl y , th is biconditional does not g ive us an a naly sis of the conc ept of a c irc le , but it is cer tainly an example of conc eptual knowledg e. I t is not analy sis of c irc le beca use wha t is on the rig ht side of the biconditional pre supposes g rasp of the conc ept of a c irc le , a nd , f ur the r, it is n ot a re su lt t ha t I a im f or in c on c e ptu a l a na ly sis . S o , al t h o u gh o n e o f t h e w ay s we ca n ac q u i re co n ce p t u al k n o wl ed ge i s t h ro u gh c o n c e p t u a l a n a l y s i s , t h e r e a r e o t h e r w a y s o f a c q u i r i n g c o n c e p t u a l k n o w l e d g e . In addition, we ca n have c onceptua l knowledg e that we know non-c onceptua lly (i.e., a posteriori). When we g ive an a ccount of what g rounds conc eptual knowledg e, then, we must be sure that our a ccount make s room for the sa me proposition to be conce ptual and non-conc eptual knowledg e. I stated above that two individuals may know the same truth, but one may know it conce ptually and the other may know it non-conc eptually (e.g ., on

PAGE 91

82 Not all scholars a g ree on who was the first to fig ure out the Py thag orea n 14 Theore m, but many seem to ag ree that Py thag oras wa s not the first. See Gorman 1979, 2, 38; Kahn 2001, 32; Neug ebaue r 1970, 36. authority or by empirica l means). Consider, f or example, the wa y that the B aby lonians fig ured out the Py thag orea n Theore m (in 1800-1600 B .C.E.) by using a tile proof: 14 Fig ure 2. Tile proof of the Py thag orea n theore m (Hig htower e t al 1997) Th e re a re e ig ht t ri a ng le s th a t ma ke up sq ua re s a and b (the leg s of the rig ht triang le, wh ic h h a s si de s f ro m sq ua re s a , b and c ). The squa re tha t is drawn extending f rom the hy potenuse of the rig ht triang le, c , contains 16 triang les. I f we add the number of triang les in the a square and b square , we g et the number of triang les in the c square . Many scholars a g ree that the B aby lonians saw this pattern of tiles to be a proof of the Py thag orea n Theore m. I t looks as if the Ba by lonians discovere d the Py thag orea n Theore m throug h empirica l means—they went out into the world and dr ew a nd counted rig ht triang les and squar es to fig ure it out. So, they knew the Py thag orea n Theore m nonconce ptually , or a poster iori. (The status of the "pr oof" as empirica l depends, of c ourse, on their taking their evide nce to c onsist of the actua l counting of the fig ures, r ather than

PAGE 92

83 any more a bstract g eometric al re asoning , but it seems at least a possible a ccount of how they took their evidenc e, and tha t suffice s for the point I wish to make.) We can a lso offer an a pr iori proof, of c ourse, of the Py thag orea n Theore m. For example, consider F igur e 3: Fig ure 3. A priori proof of the Py thag orea n theore m (Bog omolny 2005) We k no w, fi rs t of a ll, tha t th e a re a of the ins c ri be d s qu a re is c . We a lso kn ow tha t it is 2 the ar ea of the enc losing squar e minus the are as of the f our triang les forme d with the sid e s o f t he ins c ri be d s qu a re . th e le ng th o f t he sid e of the e nc los ing sq ua re is a + b , and so the a re a of the e nc los ing sq ua re is ( a + b ) . T he fo ur tr ia ng le s a re e a c h o f t he sa me 2 a re a , n a me ly 1/2 ab . S o , we h av e t h e f o l l o wi n g: c = ( a + b ) 4 x 1/2 ab 22 c = a 2 ab + b + 2 ab 22 2 c = a + b 22 2 This is the Py thag orea n Theore m, derived a priori, for it does not re st on our experience of the fig ure pr ovided as e vidence : that experience is used only by way of re prese nting the kind of fig ure w e ar e conc erne d with. Nothing hing es on the pa rticular e x perie nce, or its veridicality . None of the re asoning seems to re st in any way on a par ticular experience . So, it l ooks as if at lea st certa in ge ometrica l theorems ca n be discove red either e mpirically or conc eptually , that is, it i s possible for one to know a truth both non-

PAGE 93

84 conce ptually (as sc holars ar g ue that B aby lonians knew the Py thag orea n Theore m) and conce ptually . Th e re a re oth e r e xamp le s f ro m ma the ma tic s o f t ru ths tha t w e c a n k no w a pr ior i, but that we ca nnot know empirica lly in the way that the B aby lonians knew the Py tha g or e a n T he or e m. F or e xamp le , th e re is a pr oo f t ha t th e re is n o g re a te st p ri me nu mbe r, wh ic h g oe s a s f oll ow s: r 12 1. Sup po se tha t p =2 < p = 3 < . . . < p are all of the primes. r 12 2. L e t P = pp . . . p + 1 a nd le t p b e a p ri m e d i v i d i n g P . r 12 3. Then p can not be any of p , p , . . . , p , otherwise p would divide the r 12 differ ence P pp . . . p =1, which is impossible. r 12 4. So this prime p is stil l another pr ime, and p , p , . . . , p would not be all of the pr ime s. 5. There fore , there is no gr eate st prime number. (Ca ldwell 2005) I n p ri nc ipl e , w e c a nn ot k no w t ha t th e re is n o g re a te st p ri me nu mbe r b y a pp e a lin g to experience , since that would re quire de termining for e very number (the re a re, of course , an infinite number of numbers) w hether it is prime. Determining for e v e ry number wh e the r i t is pr ime wo uld re qu ir e a n in fi nit e a mou nt o f w or k, a nd , a s a re su lt, it c ou ld never be complete d. I ha ve c la ime d th a t c on c e ptu a l kn ow le dg e is a bo ut t he wo rl d. We s ho uld distinguish this claim from the cla im that we have to rely on particula r experienc es in the A world to ac quire c onceptua l knowledg e. Af ter a ll, conceptua l analy sis is supposed to be something tha t we ca n do independe nt of any particula r experienc es, but if conc eptual kn ow le dg e is a bo ut t he wo rl d, the n w e ha ve to a dd re ss t he iss ue of ho w i t is po ssi ble to acquir e such know ledg e without rely ing on a ny particula r experienc es. I n the ca se of de te rm ini ng wh a t c ou nts a s k no wl e dg e , n o p a rt ic ula r e xpe ri e nc e c a n te ll m e the tr uth c on dit ion s f or a pr op os iti on , f or e xamp le , n o p a rt ic ula r e xpe ri e nc e c a n te ll m e tha t I mus t have a justified, true belief that is undefea ted in order to have knowle dg e. One way we

PAGE 94

85 I will ex plain the diffe renc es and similarities betwe en scie ntific thoug ht 15 experiments and philosophical ones later in this section. can g o about fig uring out what counts as knowle dg e is by considering various, per haps imag inary , case s of what a ppear to be instance s of knowledg e and looking to see whe ther they are , in fac t, instances of know ledg e. I n the next section, I will ex amine the r ole of A thought e x periments in conc eptual ana ly sis and in ac quiring conce ptual knowledg e. Entertaining various sce narios does not re quire a ctually experiencing those scena rios or the likelihood of experiencing them. As a matter of fa ct, much of c onceptua l analy sis (or parts of c onceptua l analy ses) a mounts to entertaining various imag inary case s and re fl e c tin g on the m w he n a sk ing qu e sti on s a bo ut s itu a tio ns . I ma g ini ng su c h s c e na ri os is supposed to help us g et clea rer on things in the wor ld without directly having to look at or study what is in the world. F ram ework f or Thin kin g About Th ought Exp erim ents Th ou g ht e xpe ri me nts pla y a c e ntr a l r ole in t he me tho ds tha t ph ilo so ph e rs us e in trac ing c onceptua l connec tions and providing ne cessa ry and suff icient applica tion c on dit ion s f or ou r c on c e pts . Th ou g ht e xpe ri me nts a lso pla y a c e ntr a l r ole in 15 e sta bli sh ing ph ilo so ph ic a l c la ims or te sti ng the m, a nd we c a n lo ok a t pa rt ic ula r c a se s to see how the conc ept we a re a naly zing g ener ally applies. So, we should be c lear about w h e n w e h a v e c o n d u c t e d a t h o u g h t e x p e r i m e n t p r o p e r l y. When we conduc t thought experiments, if we do them proper ly , we a cquire conce ptual knowledg e. Conce ptual knowledg e is knowledg e that we can a rrive a t wi tho ut h a vin g a ny pa rt ic ula r e xpe ri e nc e . I t w ill e me rg e ho w t ho ug ht e xpe ri me nts he lp us arr ive at conc eptual knowledg e. B eca use thoug ht experiments play such an important role in conc eptual ana ly sis, it i s important when study ing c onceptua l analy sis to get c lear

PAGE 95

86 No t a ll t ho ug ht e xpe ri me nts a re e xplic itl y se t up lik e thi s. Ho we ve r, I thi nk tha t, 16 in p ri nc ipl e , a ny tho ug ht e xpe ri me nt c ou ld b e se t up lik e thi s. Of c ourse we conduct thoug ht experiments under an ide alization of the 17 conditions under which w e ac tually conduct them. F or example, we ig nore the fac t that many of our pr edica tes are vag ue. I use the e x pression ' scena rio intuiti on' beca use I want to make a distinction 18 betwee n intuiti ons that one may have a bout thought experiments and other kinds of philosophical intuit ions. By distinguishing the m, I do not mean to say that they have nothing in common with one a nother. on the ro le of tho ug ht e xpe ri me nts in a c qu ir ing c on c e ptu a l kn ow le dg e a nd wh y so me philosophers are skeptical a bout their utilit y . Part of g etting c lear on the role of thought e xpe ri me nts is g e tti ng c le a r o n w ho sh ou ld p a rt ic ipa te in t he m a nd wh a t c on sti tut e s a g ood thought e x periment (i.e ., one that ac tually helps us acquir e conc eptual knowledg e). The g oal of this section is to provide a f rame work for thinking about thoug ht e xpe ri me nts tha t w ill he lp u s to dis tin g uis h g oo d th ou g ht e xpe ri me nts fr om b a d o ne s in the hopes that it will eliminate some g ener al skepticism about the method of using thought e x periments to re ach me taphy sical conc lusions. Furthermor e, I hope to show that thought e x periments help us a cquire conce ptual knowledg e, and, thus, they are a va lua ble too l in do ing c on c e ptu a l a na ly sis . Se tt ing Up t he F r am e wor k: Ca noni c al Q ue st ion s a nd Sc e nar io I nt uit ion s I n th is s e c tio n, I wi ll s tip ula te de fi nit ion s o f t he c e ntr a l te rm s w e us e , o r s ho uld use, in talking about thoug ht experiments (e.g ., of ' thought e x periment' and ' intuiti on' ). A thought experime nt c on sis ts o f a po ssi ble sc e na ri o th a t a n in div idu a l is fi rs t a sk e d to imag ine and then a sked to answe r some questions about (e .g ., whether a ce rtain conc ept a pp lie s o r f a ils to a pp ly , o r w he the r i t is ind e te rm ina te tha t it do e s) . Ca ll t he po ssi ble 16,17 sc e na ri o to be c on sid e re d th e ' ta rg e t sc e na ri o' , th e re sp on se tha t th e ind ivi du a l is a sk e d to g ive a ' scena rio intuiti on' , and the r eleva nt question a ' canonic al question' . A targ et 18

PAGE 96

87 For example, see the ve ry simple thought experiment about the c oncept of a cat 19 below. Although I only list one modal question for ea ch ca nonical question, there are 20 oth e r m od a l va ri a tio ns on tho se qu e sti on s. sc e na ri o s ho uld be c omp le te ly a nd c a re fu lly de sc ri be d in the re le va nt s e ns e . A c omp le te descr iption of a targ et sce nario doe s not require a complete descr iption of the world, but only a re levantly complete de scription of wha t is relevant to the que stion that is asked. A rele vantly complete de scription of a ta rg et sce nario ne eds only to be complete in the sense that it includes a ll of the details re levant to the par ticular thoug ht experiment. For e xamp le , if a ll w e wa nt t o d o in a pa rt ic ula r t ho ug ht e xpe ri me nt i s w e e d o ut c e rt a in conditions as being nece ssary and/or suff icient for the application of a conce pt, then a rele vantly complete de scription need not c ontain as many details as a thoug ht experiment involving the sa me conc ept the g oal of whic h is to establish necessa ry and suff icient conditions. A ca ref ul description of a targ et sce nario should not include irre levant or 19 extra details that might conf use or influe nce a participa nt's intuitions. Th e re a re ty pic a lly thr e e kin ds of c a no nic a l qu e sti on s o ne ma y a sk a pa rt ic ipa nt i n a thoug ht experiment to answer: (a ) metac onceptua l questions, for example, "D oes the conce pt of knowledge apply in this si tuation? "; ( b) metaling uistic questions, for example, "D oes the re lational term ' knows' rela te A and p in this situation? "; a nd (c) first-orde r questions, for example, " Does A know that p? " T here are also modal varia tions on canonic al questions (a) –(c) : (a' ) " I n such a situation, could the c oncept of knowledge 20 apply ? "; ( b' ) " I n such a situation, could the r elational ter m 'know s' apply to the pair of A a nd p?"; a nd (c ' ) " I n s uc h a sit ua tio n, c ou ld A kn ow tha t p?" Ad dit ion a lly , w e c a n a sk canonic al questions that have a ux iliary premises (e .g ., "Pre suming that the c oncept of

PAGE 97

88 Be low, I will give e x amples of thoug ht experiments that use auxi liary premises 21 to show this more clea rly . Canonical que stions that are a ccompa nied by auxil iary pr e mis e s c a n c on ta in m od a l te rm s. ba ldn e ss is a conc ept with deter minate applica tion conditions, does the conce pt of ba ldn e ss apply in this si tuation? ") . As I stated above , the re sponse one g ives on the 21 ba sis of ho w i t se e ms t o h im t o a ny of the se qu e sti on s is a sc e na ri o in tui tio n, bu t le t us , m o re p re ci s el y , s ay t h e f o l l o wi n g: A sc e na rio intuiti on is a n u nme dia te d c og nit ive se e min g a bo ut a ta rg e t sc e na ri o that could serve as the ba sis of an answe r to a que stion about the targ et sce nario. A s c e na ri o in tui tio n is a re sp on se to a ta rg e t sc e na ri o th a t, l ike a be lie f, ha s c on te nt a nd is not just a first impression, and, unlike a be lief, is not something ne cessa rily endorse d by the individual responding to the targ et sce nario. A sc enar io intuit ion must be unmediated in t he se ns e tha t no the or y sh ou ld b e op e ra tin g in t he ba c kg ro un d. A s c e na ri o in tui tio n is only a spec ies of philosophical intuition, and I do not mean to cove r all of the dif fer ent kinds of philosophical intuit ions in my char acte rization. For my purposes in this section, I am only interested in our intuiti ons about thoug ht experiments. Ty pic a lly , w he n w e c on du c t th ou g ht e xpe ri me nts , w e a re su pp os e d to g a in c on c e ptu a l kn ow le dg e (t ha t is , k no wl e dg e of c on c e ptu a l tr uth s, wh ic h a re ne c e ssa ri ly tr ue ). Whe n w e c on du c t th ou g ht e xpe ri me nts , th e n, we sh ou ld w a nt r e sp on se s to canonic al questions that enable us to come to know something modal in char acte r. Howeve r, althoug h I want to flag the appe ara nce of modality , as I do when I list t he canonic al questions, I do not want to build modality into the notion of scenar io intuit ion itself beca use there are three way s that modality may show up in conducting thought experiments, two way s that are rela ted to the par ticipant and one w ay that is related to the experimenter. Since the participa nt and not the experimenter is supposed to g ive us

PAGE 98

89 This targ et sce nario, a lthough it does not conta in many details, display s the 22 re le va nt c omp le te ne ss r e la tiv e to i ts g oa l. intuiti ons, it would not m ake se nse to include talk a bout modality in the definition of ' sc e na ri o in tui tio n' . I f t he e xpe ri me nte r a sk s a qu e sti on tha t c on ta ins a mod a l te rm in i t, the n th e sc e na ri o in tui tio n th a t th e pa rt ic ipa nt g ive s w ill no t be mod a l in c ha ra c te r. B ut i f the no tio n o f s c e na ri o in tui tio n h a s mo da lit y bu ilt int o it , th e n th a t ki nd of int uit ion wo uld not count as a sc enar io intuit ion. This will become clea rer below. First, I have a lrea dy said that we c an ask the participa nt a ca nonical question that has a moda l term in it. That is, the ex perimente r prompts the par ticipant by loading the question with a modal term, and the participa nt is will ing to g ive an intuition involvi ng an explicit modal noti on. L et us ca ll this a prompted moda l response. Sec ond, the pa rt ic ipa nt h e rs e lf ma y g ive a re sp on se tha t ha s mo da l c on te nt. L e t us c a ll t his a spontaneous modal re sponse. Third, the e x perimente r dra ws a modal c onclusion from the par ticipant' s response to a question that does not have moda l content. I n this case, the experimenter sa y s something modal that doe s not depend solely on what the pa rticipant sa y s, bu t on ho w t he e xpe ri me nte r u nd e rs ta nd s th e pa rt ic ipa nts ' int uit ion s. L e t us c a ll this the ex perimente r' s modal infere nce, or modal infere nce, f or short. To see how the three differ ent way s modality may enter the picture , let us consider diff ere nt way s that mod a lit y ma y a pp e a r i n a ve ry sim ple tho ug ht e xpe ri me nt: (1 ) S up po se tha t th e re is a c re a tur e tha t me ow s, ha s f ou r l e g s, wh isk e rs , a ta il, he a rt , a nd lun g s, bu t no fu r. Cou ld t his c re a tur e be a c a t? 22 Assuming that the pa rticipant answe rs " y es" to the canonic al question, we c an conc lude from this (albeit ver y simple) targ et sce nario that the pa rticipant thinks that it is not nece ssary for a cat to have fur ( i.e., there can be furless c ats). The participa nt's intuition

PAGE 99

90 has modal conte nt beca use it is in an answer to a question that contains a moda l term—she g ave a prompted modal re sponse. So, having fur is not a ne cessa ry condition for be ing a cat. This thoug ht experiment give s a desc ription of a cr eatur e that appe ars very much to be a c at, aside f rom one fe ature , having fur. This would be a g ood thought experiment beca use it clea rly g ets at the intuitions it intends to, namely , intuiti ons about our conc ept of a cat , and it isolates just one fea ture that is in contention for be ing a n application condition of the c oncept. (2 ) S up po se tha t th e re is a c re a tur e tha t me ow s, ha s f ou r l e g s, wh isk e rs , a ta il, he a rt a nd lun g s, bu t no fu r. I s th is c re a tur e a c a t? Consider now the sce nario a nd question in (2). Assuming that the par ticipant answe rs " it' s p os sib le " to t he c a no nic a l qu e sti on a nd is t ry ing to b e a s in fo rm a tiv e a s p os sib le , it is clea r that the pa rticipant thinks that the targ et sce nario doe s not give enoug h information about the cr eatur e; it is open relative to the de scription that it is a cat or it is not a ca t. That is, it may be that the pa rticipant thinks that all of the fe ature s listed are ne cessa ry for application of the conce pt of cat , but they are not sufficient. She may think, on the other hand, that none a re inc ompatible with being a ca t (that is, none ar e nec essar y ), but none rule out its being a ca t. Or, she may think that some of the fe ature s listed are ne cessa ry for be ing a cat, but they are not jointl y sufficie nt. I n any case , the par ticipant' s intuit ion has modal conte nt—she g ives a spontane ous modal response. L et us assume that the pa rticipant answe rs " y es" to (2). Ne ither the c anonica l question nor the par ticipant' s response c ontains a modal term. How ever , modality may enter the picture when the e x perimente r eva luates the tar g et sce nario. Suppose that the experimenter is intere sted in fig uring out whether having fur is a ne cessa ry condition for being a ca t. After g ather ing se vera l participants' intuiti ons of the above form, she ma y

PAGE 100

91 Sometimes there may be no straig htforwa rd answe r to ca nonical questions 23 be c a us e the re is a pr e su pp os iti on tha t pr e dic a te s a re no t va g ue wh e n th e y a re , o r t he re is not enoug h information in the targ et sce nario. I n case s of vag ue pre dicates, the rules for application of the predic ate g ive out (i.e., they do not tell us what to say ). By ' did' he re D avidson does not nec essar ily mean that the pe rson per formed a n 24 ac t i o n . He d ra ws a d i s t i n ct i o n b et we en 'd o i n g' an d 'ac t i n g': "W ak i n g u p i s s o m et h i n g I di d, perh aps , b ut no t an act io n" (Da vidson 1980, 43). infer tha t having fur is not a ne cessa ry condition for being a ca t—it is possible for a c re a tur e tha t la c ks fu r t o b e a c a t. T his is a sit ua tio n in wh ic h th e e xpe ri me nte r m a ke s a modal judgme nt based on something without modal content. This is a modal infere nce made by the experimenter. Clearly the ca ses we or dinarily consider in doing philosophy are not as simple as the one I descr ibed above . L et us now consider a more complicated e x ample fr om Da vid so n, wh ic h in vo lve s a c a no nic a l qu e sti on c on ta ini ng a c ou nte rf a c tua l: 23 This morning I was a wake ned by the sound of someone prac tising the violin. I dozed a bit, then g ot up, washed, sha ved, dre ssed, and we nt downstairs, turning off a light in the hall as I passed. I poured my self some c offe e, stumbling on the edg e of the dining r oom rug , and spilled my coff ee f umbling for the Ne w York Times . . . . I might have turned off the light by inadver tently brushing ag ainst the switch; would it then have be en my deed, or even some thing that I did? 24 (Da vidson 1980, 43) The c onclusion that Davidson re ache s is that turning off the light in this case was so me thi ng tha t he did —a ft e r a ll, wh e n w e ta lk a bo ut doing something ina dverte ntly , we us e a n a c tio n v e rb . B ut, thi s r a ise s a fu rt he r q ue sti on a bo ut w he the r h e did it intentionally . On Davidson' s view, an a dverb is alwa y s relativized to a descr iption—under one de scription it may be that a pa rticular a ction was pe rfor med intentionally , under a nother unintentionally , and, under a third, it may be open w hether the ac tion was intentional or not. I f we focus on the la tter question, it seems that

PAGE 101

92 This is the problem that comes up in Wilkes's (1988) case , which I discuss in the 25 section of Chapte r 3 entitled, " Sources of Error and Corre ction." a lth ou g h tu rn ing on the lig ht w a s so me thi ng he did , it wa s n ot s ome thi ng he did intentionally . To see how one g ets this intui tion, we nee d to unpack the te rm ' inadver tently ' . When we do something inadver tently , we do it without rea liz ing tha t we are doing it, and it see ms that we do it acc identally . Doing something w ithout realizing we a re doing it and doing it by acc ident just does not seem to suffice for doing something intentionally . I f we conce ntrate on the latter que stion about whether the action wa s done int e nti on a lly , o nly on e fe a tur e of a c tio n is be ing c a lle d in to q ue sti on in t his po ssi ble scena rio, and ba sed on one' s response to the que stion invol ving tha t fea ture, we may infer something more g ener ally about what the pa rticipant in the thoug ht experiment thinks about the conc ept of action —though doing something ina dverte ntly is an ac tion, it i s not an intentional one. I n other wor ds, the conditions describe d in the targ et sce nario a re sufficie nt for the turning on of the lig ht to be an ac tion, but not s ufficie nt for the turning on of the lig ht to be intentional. Th e c a no nic a l qu e sti on a sk e d a t th e e nd of the de sc ri pti on of the ta rg e t sc e na ri o contains a c ounterfa ctual. Commonly , althoug h not in this case , the counter fac tual question is of the form, " What would we say if . . .? " T he question should be under stood as asking what our intuitions tell us would be or must be true in the re levant ca se. When answe ring canonic al questions that contain counte rfa ctuals that we should make sure we keep the conce pt we ar e conside ring fix ed, or a t least we be car eful a bout counterf actua l questions that may not keep the c oncept f ix ed. Fur thermore , we ha ve to kee p in mind 25 that are asking about whethe r our c oncepts a pply to the targ et sce nario, not whe ther the individual's (in the ta rg et sce nario) c oncept a pplies.

PAGE 102

93 Whe n ta lki ng a bo ut i ntu iti on s a nd the e xten t to wh ic h th e y ha ve mod a l c on te nt, 26 we should be c lear about the diffe renc e betwe en the a lethic, or ordina ry , notion of modality and the e pistemic notion of modality . For example, consider the f ollowing qu e sti on : D id t he pe rs on do it i nte nti on a lly ? I f a pa rt ic ipa nt a ns we rs , " No , n e c e ssa ri ly the per son did not do it intentionally ," w e should understa nd the ' nece ssarily ' in the alethic se nse (i.e., in the se nse that whe ther she ha s the corr ect intuition or not, the corr ect a nswer to the que stion is s omething tha t is necessa rily the ca se, and not in the epistemic sense ). We ca n understand the epistemic sense of possibilit y and nec essity as fo llo ws : That p is epistemically possible for x iff that p is compossibl e with the p ro p o s i t i o n s x k n o ws o r i s j u s t i fi ed i n b el i ev i n g. That p is epistemically nece ssary for x iff it is not epistemically possible that not p. I n other wor ds, we do not wa nt to say that the par ticipant' s intuit ions follow from any thing e lse she knows that the pe rson did it intentionally . I t follows from the descr iption of the situation that she has not done it intentionally . Epistemic nece ssity has to do with what we know a bout the actua l world, and since thought e x periments ar e not about the ac tual world, epistemic possibility and nec essity are not releva nt. Thoug ht experiments have to do with what we know about eve ry possible world. The a nswer tha t a par ticipant g ives to such a que stion may not contain a modal term, in which c ase the theorist may make a modal infere nce. A t the end of this targ et scena rio, Davidson (1980) could have asked e ither of the f ollowing que stions that have modal content: (a ) " I n such a situation, is it necessa ry that he per formed a n action? ," or (b) " I s it possi ble for some one to do all of these things a nd it not be an ac tion? " T he response that a par ticipant would g ive to either of these que stions would be a prompted mo dal res po ns e. L et u s l oo k at a si mp le t ho ught ex peri men t f rom Aus ti n's (1 95 6) "A 26 Plea for E x cuses," which also involves the c oncept of action but which distinguishes doing some thing by acc ident and doing something by mistake: You have a donkey , so have I , and they g raze in the same field. The da y comes when I conce ive a dislike for mine. I g o to shoot it , draw a bea d on it, fire: the br ute fa lls in i ts t ra c ks . I ins pe c t th e vic tim , a nd fi nd to m y ho rr or tha t it is your do nk e y . I a pp e a r o n y ou r d oo rs te p w ith the re ma ins a nd sa y —w ha t? 'I sa y , o ld s p o r t , I' m a w f u l l y s o r r y, & c . , I' v e s h o t yo u r d o n k e y by acc ident ' ? Or ' by mistake ' ? Th e n a g a in, I g o to sh oo t my do nk e y a s b e fo re , d ra w a be a d o n it , f ir e —b ut a s I do so, the bea sts move, and to my horror y ours fa lls. Aga in, the scene on the d o o r s t e p — w h a t d o I s a y ? ' B y m i s t a k e ' ? O r ' b y a c c i d e n t ' ? (Austin 1956, 185)

PAGE 103

94 I ntere stingly , Tamar Gendler (1998, 2000) doe s not in any of her work on 27 tho ug ht e xpe ri me nts g ive a c ha ra c te ri zat ion of int uit ion s. B e a le r d ra ws a dis tin c tio n b e tw e e n tw o k ind s o f i ntu iti on s, ra tio na l in tui tio ns , 28 which we use in philosophy and phy sical intuitions, which we use in the scie nces. Be al er 's 'ra t i o n al i n t u i t i o n s ' ar e w h at I re fe r t o as 'p h i l o s o p h i ca l i n t u i t i o n s ' (B eale r 1998a) . After eac h targ et sce nario, the pa rticipant is asked a first-orde r question that does not have moda l content. I t seems that the intuition mos t participants would have is, in the first scena rio, " by mistake" or " definitely by mistake," wher e ' definitely ' has modal for ce. I n the sec ond scena rio, it seems that a pa rticipant would say either " by acc ident" or "de finitely by acc ident." T his thought experiment shows that there is a distinction betwee n doing some thing by mistake and doing it by acc ident beca use we have be en g iven two similar (but distinct) targ et sce narios about whic h we would sa y that in one of the m so me thi ng wa s d on e by mis ta ke a nd in a no the r s ome thi ng wa s d on e by a c c ide nt, but not vice ver sa. The notion of intuition that I have de scribed in this cha pter is limited to responses to t ho ug ht e xpe ri me nts , b ut ' int uit ion ' c a n b e a pp lie d mo re br oa dly . A t th is p oin t, w e wi ll re vie w h ow oth e rs c ha ra c te ri ze i ntu iti on s. B e a le r c ha ra c te ri zes the m a s f oll ow s: 27 For y ou to have [a philosophical] int uition that A is j ust for it to seem to y ou that A. . . . F or e xamp le , w he n y ou fi rs t c on sid e r o ne of de Mo rg a n' s la ws , o ft e n it neither se ems to be true nor seems to be f alse; af ter a moment's r efle ction, ho we ve r, so me thi ng ha pp e ns : it no w s e e ms t ru e ; y ou su dd e nly " jus t se e " tha t it is tr ue . O f c ou rs e , th is k ind of se e min g is intellectual , not sensory or introspec tive. (B eale r 1996, 123) 28 One of the obvious differ ence s betwee n the way that Be aler char acte rizes intuit ions and the wa y I do is t ha t he do e s n ot l imi t in tui tio ns to r e a c tio ns to t ho ug ht e xpe ri me nts ; intuiti ons can be any kind of intellectual see ming ( e.g ., intellectual see mings a bout mathematica l theorems or la ws of log ic). I n addition, I have le ft it open whethe r

PAGE 104

95 modality appea rs in intuiti ons, where as B eale r cla ims that all of our intuiti ons have modal content. F or example, on Be aler ' s view, a pa rticipant in a thoug ht experiment would either g ive a spontane ous or a pr ompted modal response to a ca nonical question. T h a t i s , B e a l e r t h i n k s t h a t i f I i n t u i t t h a t x , i t s e e m s t o m e t h a t x i s n e c e s s a r y: [W]he n w e ha ve [a p hil os op hic a l] in tui tio n, sa y , th a t if P th e n n ot n ot P , th is prese nts itself as nece ssary : it does not seem to us that things could be othe rwise; it must be that if P then not not P. (B eale r 1992, 102) There are two possible rea dings of this, and Be aler does not make it cle ar w hich one he intends. The most natural r eading is that Bea ler thinks that all of our intuitions have modal content. Another rea ding of it is t hat we g et two intuitions at once : (a) if P then not not P and (b) nec essar ily , if P then not not P . Whether we understand it the f irst way or the se c on d, my vie w i s w e a ke r. I ha ve c la ime d th a t in tui tio ns do no t have to have mod a l c on te nt a t a ll; it m a y be tha t th e e xpe ri me nte r m a ke s a mod a l in fe re nc e fr om a participa nt's intuitions. Bea ler' s char acte rization of philosophi cal intuition is not inconsistent with the one I have pr ovided, thoug h in one sense it is broade r than mine, since it cove rs all intellectual se eming s where mine only cover s those seeming s we g et as a re sult of considering a tar g et sce nario, a nd in another it is more na rrow, sinc e on Be aler ' s view modal notions have to appe ar in the intuition itself. That is, it seems that on Be aler ' s view our intuitions are spontane ous modal responses. L ike B eale r, I have c alled sce nario intuitions 'unmediate d cog nitive seeming s' . I do not call them ' judgme nts' be cause I want to make r ooms for par ticipants having the fol lo wi ng in tu it io ns abo ut tar get s cenar io s: "I t s eem s t o m e th at x , b ut I do n't b eli eve i t. " Consider, for e x ample, Chalmers' s zom bie thoug ht experiment where an individual who is phy sically identical with us is describe d in such a wa y that it appear s that it does not ha ve c on sc iou s e xpe ri e nc e s (Chalmers 1996, 94f f). Par ticipants in those kinds of thought

PAGE 105

96 See also 29 Be aler 1996. experiments should be able to say , "I t seems to me that the individual lacks c onscious experience , but I do not believe that it does." I take it that judg ments are a spec ies of belief, a nd, so, I do not want intuitions to be judg ments. Howeve r, this is not t o say that there is no connec tion whatsoever betwee n seeming s and belief s: a seeming could be g ood evidenc e for believing something. Tha t is, there ar e cir cumstance s in which I have a n u nme dia te d c og nit ive se e min g tha t se rv e s a s e vid e nc e fo r m y be lie vin g wh a t se e ms t o me to be the c ase. See mings c an have independe nt cre dibilit y . Simi larly , Be aler arg ues tha t in tui tio ns a re no t be lie fs , a nd , th us , a re no t ju dg me nts : [T] here are many mathematica l theorems that I believe ( beca use I have se en the pr oo fs ) b ut t ha t do no t seem to m e to b e tr ue a nd tha t do no t seem to me to be false; I do not have intuitions about them either wa y . Conversely , I have a n int uit ion —it sti ll seem s to me—that the naive c omprehe nsion ax iom of set theory is true; this is s o despite the fa ct that I do not believe that it is true (be cause I kn ow of the se tthe or e tic a l pa ra do xes) . T he re is a ra the r s imi la r p he no me no n in se ns e pe rc e pti on . I n th e Mü lle rL y e r i llu sio n, it s til l se e ms to me that one of the two arr ows is longe r than the othe r; this is so despit e the f act that I do not believe that one of the two arr ows is longe r (be cause I have me asure d them). I n eac h case , the see ming pe rsists in spi te of the c ounterva iling belief . . . . I ntuitions are also quite distinct from judgme nts, gue sses and hunc hes. . . . J udg ments are a kind of oc curr ent belief ; as such, they are not seeming s. (B eale r 1992, 102-103) 29 Wha t I ha ve sa id a bo ut i ntu iti on s is ne utr a l a bo ut c a se s in vo lvi ng ma the ma tic a l th e or e ms since they do not involve targ et sce narios. I n Intuitions as Evidence , Pust (2000) char acte rizes intuit ions in a way very simil ar to B eale r (1992, 1996) : S ha s a [phi los op hic a l] in tui tio n th a t p I F AN D O NL Y I F (a ) S ha s a pu re ly intellectual experienc e, whe n considering the question of whe ther p, that p; and (b) a t t, if S were to conside r whe ther p is nec essar ily true, then S would have a purely intellectual experienc e that nec essar ily p. (39)

PAGE 106

97 This is a slightly wea ker c hara cter ization t han B eale r' s notion of intuit ion beca use it does not require that our intuitions have modal content. I t makes room for prompted modal response s to targ et sce narios. F or example, if someone ha d the intuition t hat p and we re a sk e d if sh e tho ug ht i t w a s n e c e ssa ry tha t p, the n s he wo uld re sp on d " y e s. " Simi la rl y to Be al er , P u s t d o es n o t al l o w f o r t h e e x p er i m en t er t o m ak e a n i n fe re n ce fr o m a p ar t i ci p an t 's intuiti on that has modal conte nt, (i.e., there is no room for infe renc e-dr iven re sponses by the experimenter) . Although Pust is silent about whether be lief is sufficie nt for intuition, he does ag ree with Be aler (1992, 1996) ( and my self) tha t beliefs ar e not nec essar y for intuition: . . . I think attention to any philosophical para dox is sufficient to show that belief is not necessa ry for intuition. I n a par adox, one is confronted w ith intui tive pr op os iti on s w hic h le a d b y int uit ive ly a c c e pta ble re a so nin g to a n in tui tiv e ly unacc eptable c onclusion. After noting that one is presente d with a par adox, one is l ike ly to s us pe nd be lie f i n a ll t he re le va nt p ro po sit ion s w ith ou t th e ir diminis hing in intuitive forc e. F urthermor e, eve n afte r one ha s come to a resolution of the pa radox and ac tually disbelieves one of the propositions, the offe nding pr oposition often retains a f air a mount of intuit ive plausibility . On the streng th of these sorts of e x amples, I conclude that belief is not nec essar y for intuiti on. (Pust 2000, 33-34) L e t us no w t a ke a loo k a t So sa ' s ( 19 98 ) c ha ra c te ri zat ion of a n in tui tio n, wh ic h a c tua lly bu ild s in the no tio n o f b e lie f: At t, i t is int uit ive to S tha t p i ff (a ) i f a t t S w e re me re ly to u nd e rs ta nd fu lly enoug h the proposition that p (absent r eleva nt perc eption, introspection and rea soning) , then S would believe that p; (b) at t, S does [full y ] understand the proposition that p; (c) the pr oposition t hat p is abstrac t. (Sosa 1998, 259) Sosa' s char acte rization of intuit ion includes belief but only on the condition that the individual actually understands the proposition in question. Or, minim ally , it does not se e m th a t on e wh o d oe s n ot u nd e rs ta nd p w ou ld h a ve a re lia ble int uit ion . I wi ll d isc us s this in detail in t he following section. Sosa' s inclusion of (c) in his account of

PAGE 107

98 so me thi ng ' s b e ing int uit ive is i nte re sti ng be c a us e it s e e ms t o h a ve so me c on ne c tio n w ith the issue of modality . I side with Be aler (1992, 1996) a nd Pust (2000) about intuiti ons being cog nitive se e min g s b e c a us e it s e e ms p os sib le fo r u s to be un su re a bo ut o ur int uit ion s, a nd in t ho se circ umstances we would not believe wha t we intuit (simil ar to c ases of perc eptual ill us ion ). No ne the le ss, se e min g s a re be lie flik e in t ha t th e y a re pr op os iti on a l a tti tud e s, that is, in S ear le' s terms, they have mind-w orld direc tion of fit and have a conte nt char acte rized by a proposition. This fac t is im portant for understanding how intuitions can suppor t beliefs about ne cessity and possibility and about a pplication conditions for conce pts. That is, although sc enar io intuit ions are not be liefs, they seem to provide support for them. So, they play the same r ole that per ceptions play in justi fy ing pe rc e ptu a l be lie fs . Scientifi c and Philosoph ical Thou ght Experim ents Just a s th e re is a me tho d f or do ing e xpe ri me nts in s c ie nc e , th e re is, or a t le a st oug ht to be, a method for conducting thought e x periments, both in philosophy and scienc e, since they are experiments of a c erta in kind. Once we g et clea r on the rule s for c on du c tin g tho ug ht e xpe ri me nts , w e wi ll b e a ble to l oo k a t va ri ou s th ou g ht e xpe ri me nts and dete rmine which one s are g ood ones (that is, which one s obey the rules) and which ones ar e bad one s (that is, ones which do not obey the rules) . I will arg ue below tha t those who ar e skeptica l about the utility of thoug ht experiments often use bad thoug ht experiments, or use g ood thought e x periments in the wr ong way in arg uing one should be sk e pti c a l. S ur e ly we wo uld ha ve re a so n to be sk e pti c a l of tho ug ht e xpe ri me nts on ly if things we nt wrong even w hen we followed a ll the rules and use g ood thought experiments. This has never been show n.

PAGE 108

99 I should point out Be aler that thinks that only scientific thoug ht experiments are 30 actua lly thought e x periments. I nstead of c arving up thought e x periments in the wa y I do, Be aler car ves up two diffe rent notions of intuition (B eale r 1998a, 207208). For a more detailed discussion of sc ientific thoug ht experiments, see 31 Br own 1991. B e fo re g ivi ng the c on dit ion s f or pr op e rl y pe rf or min g tho ug ht e xpe ri me nts in philosophy , we should g et clea r on the diff ere nce be tween a philosophical thought e xpe ri me nt a nd a sc ie nti fi c tho ug ht e xpe ri me nt. A p hil os op hic a l th ou g ht e xpe ri me nt i s 30 de sig ne d to te st w he the r a pa rt ic ula r c on c e pt w e po sse ss i s a pp lic a ble to a ta rg e t sc e na ri o and to establish something about its application conditions, which will have implications fo r w ha t c on c e pt a c tua lly a pp lie s. A s c ie nti fi c tho ug ht e xpe ri me nt h old s a sc ie nti fi c theory fix ed and te sts to see what the the ory is committ ed to and wha t the conseque nces of it are g iven a possible situation. That is, in the descr iption of the case , the e x p e r i m e n t e r s a ys , " A s s u m i n g t h e t h e o r y T is tr ue, th en w hat wou ld hap pen if . . .? " Th e re is n o s uc h c on str a int on ph ilo so ph ic a l th ou g ht e xpe ri me nts . Ra the r, in 31 ph ilo so ph ic a l th ou g ht e xpe ri me nts we te st a the or y to s e e wh e the r i t is c on sis te nt w ith what we think is t rue a bout the world indepe ndent of the the ory , that is, by asking for a participa nt's pr etheor etic intuitions about certain que stions reg arding a tar g et sce nario. That is, the par ticipant' s pretheor etic intuitions are supposed to g ive us information about wh a t is a le thi c a lly ne c e ssa ry a nd po ssi ble . T he n w e loo k a t th e the or y to s e e wh e the r i t matches up w ith the participant' s intuit ions. Of cour se, one ma y arg ue that wha t we believe a bout the world is not going to tell us much about what is true a bout the world, since sure ly we ha ve fa lse beliefs. H oweve r, wha t we ar e doing when we conduct t h o u gh t ex p er i m en t s i s co n ce p t u al an al y s i s ( i . e. , we ar e a n al y z i n g ou r conce pts). I f we g e t th e ri g ht p a rt ic ipa nts , it is d if fi c ult to s e e ho w w e c ou ld g e t a na ly se s o f o ur c on c e pts

PAGE 109

100 wr on g (a ssu min g we do e ve ry thi ng e lse c or re c tly , o f c ou rs e , a nd a ssu min g a n e pis te mic view of c oncept possession, as I do). To se e mor e c le a rl y the dif fe re nc e s b e tw e e n p hil os op hic a l th ou g ht e xpe ri me nts and scie ntific thoug ht experiments, let us first consider the f ollowing thoug ht experiment introduced by Wil liams (1970) in" The Self a nd the Futur e." Then we will compare it to a sc ie nti fi c tho ug ht e xpe ri me nt: Two per sons, person AA , th e pe rs on a sso c ia te d w ith ps y c ho log ic a lsta te sA , body /brainA and per son BB , th e pe rs on a sso c ia te d w ith ps y c ho log ic a lsta te sB , 01 body /brainB , a re tol d a t t a bo ut a n o pe ra tio n th e y wi ll u nd e rg o a t t the re sult of 2 wh ic h is tha t a t t brainB is a sso c ia te d w ith the ps y c ho log ic a l st a te s p re vio us ly associate d with AA , and bra inA is associated w ith the psy cholog ical states previously associate d with BB . Two per sons emerg e— AB and BA . I mag ine 0, fu rt he r t ha t a t t befor e AA underg oes the proc edure , she is told that either AB or BA will be g iven a lar g e sum of money and the other will be tortured. AA is then asked to make a selfinterested de cision as to whether the re war d should be g iven 2 to AB or BA a t t . Ca noni c al q ue st ion : What would it be rational for AA to choose? I n this thought experiment, we a re g iven a sc enar io and aske d to make a judg ment a bo ut w ha t AA would rationally choose. The judgme nt we make is supposed to revea l what we think is necessar y and suff icient for pe rsonal identity , even thoug h the intuitions we a re a sked to g ive ar e not direc tly rela ted to the conc ept. The pa rticipant is asked w hat AA would rationally choose in the c ase, not who AA thinks she is identical with. Howeve r, fr om our answe r to the question we a re suppose d to be able to infe r who AA would rationally think she is identical with. I f the pa rticipant say s that it would be rational for AA to give AB the money and BA the torture ( as it seems we would presume) , since AA is supposed to make a pur ely self-inter ested de cision, then we c an infe r that the participa nt thinks t hat psy cholog ical continuity is nece ssary and suff icient for pe rsonal i d e n t i t y.

PAGE 110

101 Now let us consider Galileo' s famous thoug ht experiment in Dialogues Concerning the Two New Science s about fa lling bodies, whic h is supposed to show that Ar ist otl e ' s th e or y of the na tur e of mot ion , a c c or din g to w hic h th e we ig ht o f a n o bje c t is proportional to the spee d at which it fa lls, is i ncorr ect, or at best, incomplete (Ga lilei 16 38 /19 14 ). Th e tho ug ht e xpe ri me nt i s a s f oll ow s: I mag ine that two objects (ma de of the same mate rial), whic h differ in weig ht are strapped to one another and droppe d from a c erta in heig ht. L et A be the heavie r object and B be the lig hter objec t. Ca noni c al q ue st ion : What account would Ar istotl e (or an Ar istotl ean) g ive of t h e s p e e d a t w h i c h A + B s h o u l d f a l l ? There seem to be two thing s that the Aristotelian could say , depending on whether A+ B a nd C ( the c omp os ite ob je c t ma de of A a nd B ) a re dis tin c t ob je c ts. No ma tte r w ha t, ho we ve r, the Ar ist ote lia n is c omm itt e d to e ith e r a c on tr a dic tio n o r s ome thi ng tha t is empirica lly false: (a) The Ar istotelian might say that B w ould slow down the rate at which the A falls, and so the stra pped-tog ether object, A+ B, w ould fall slower tha n A. Or he might say that A would spee d up the ra te at which B falls, and so A+ B f alls faster than A. I t seems that the Ar istotelian has to say both that A+B falls fa ster and slower than A . We have a contra diction. So, it must be that the Aristotlean theory is incorre ct, or minimally , it is i ncomplete. (b ) A lte rn a tiv e ly , th e Ar ist ote lia n mi g ht s a y tha t A + B a nd C a re dis tin c t ob je c ts, so tha t th e c omp os ite ob je c t, C , f a lls fa ste r t ha n e ith e r A or B a lon e , s inc e it i s heavie r. Howe ver, g iven the conc lusion reac hed in (a ), the Ar istotelian has to say that A+B and C are distinct objects, and thus, that when C comes into existence, the object A+ B g oes out of existence. The Aristotelian has not spelled out wha t the la w i s su pp os e d to be (i .e ., he do e s n ot s a y e no ug h a bo ut w ha t ob je c ts a re so the ac count does not extend to combinations of objects). So, the Aristotelian theory is, at best, incomplete and unr efle ctive. Bo t h W i l l i am s 's an d Ga l i l eo 's t h o u gh t ex p er i m en t s ar e, i n a s en s e, t es t i n g a theory , the psy cholog ical continuity theory of per sonal identity and Ar istotl e' s theory of the na tur e of mot ion , r e sp e c tiv e ly . H ow e ve r, in t he sc ie nti fi c tho ug ht e xpe ri me nt, we a sk

PAGE 111

102 so me on e wh o a lr e a dy ha s a the or y a bo ut s ome thi ng to r e sp on d to the tho ug ht e xpe ri me nt, a nd in t he ph ilo so ph ic a l th ou g ht e xpe ri me nt, we a re su pp os e d to a sk so me on e wh os e response to the canonic al question will not be mediated by a theor y about per sonal identity . That is, in a philosophical thoug ht experiment, we ar e supposed to be g etting clea r on our c oncepts, a nd that only require s pretheor etic intuitions, not a theor y . The theory comes in only when we want to compar e it with the results of the thoug ht experiment. This is not t o say , of cour se, that the e x perimente r ca nnot have a theory a bo ut t he c on c e pt i n q ue sti on . I f s ome on e a llo ws he r t he or y a bo ut p e rs on a l id e nti ty to g uide her intuiti ons, then presuma bly she is not a suitable ca ndidate for participa tion in a thought e x periment be cause her se eming is not an unmediated one . But, a s I stated ear lier, it seems that the philosopher should be a ble to consider the thought e x periment independe nt of her the ory . Some philosophers might think that philosophical thought e xpe ri me nts a re ba d s c ie nti fi c tho ug ht e xpe ri me nts a nd re ly on the or ie s. B ut t his is confuse d, since in scientific thought e x periments a the ory is built into the thought e xpe ri me nt, a nd a the or y is n ot b uil t in to p hil os op hic a l th ou g ht e xpe ri me nts . Good P hilos ophi cal Thou ght Experim ents What makes a philosophical thoug ht experiment a g ood one, that is, one that we c a n u se in c on c e ptu a l a na ly sis ? At thi s p oin t in the dis se rt a tio n, we ma y le a ve sc ie nti fi c thought e x periments behind a nd continue our discussion of philosophical thoug ht e xpe ri me nts . A g oo d th ou g ht e xpe ri me nt, of c ou rs e , is g oin g to b e on e tha t a c tua lly helps us g et clea rer on the conc ept that is in question. W e ca n only g et clea rer on the conce pt in question by means of doing a thoug ht experiment if we ac tually g et intuitions from the pa rticipant. I f a thoug ht experiment does not actually g et at our intuitions, for example, if it presupposes that we g et ce rtain intuitions or if it does not give us intuitions

PAGE 112

103 about the re levant conc ept, then the thoug ht experiment will not serve its purpose. So, one condition that a g ood thought e x periment must meet is that it must get at our int uit ion s. I n th is s e c tio n I wi ll g ive e xamp le s o f g oo d th ou g ht e xpe ri me nts a nd e xpla in why they are g ood. On e e xamp le of a g oo d th ou g ht e xpe ri me nt i n th e ph ilo so ph ic a l li te ra tur e is Gettier' s Smi th and J ones example, which I explained ear lier in the sec tion of this chapter entitled, "Conc eptual Ana ly sis and Conceptual Know ledg e." This is int uitively a g ood tho ug ht e xpe ri me nt b e c a us e the de sc ri pti on of the c a se c on ta ins a ll o f t he re le va nt d e ta ils fo r t he pa rt ic ipa nt t o r e lia bly e va lua te it. Th e re is n oth ing in t he de sc ri pti on tha t is misleading to the participa nt. The re sult of the thought e x periment is that it shows convincing ly that the traditional ana ly sis of the conc ept of knowledge is, a t be st, incomplete. One who fully possesses the c oncept of knowledge will have the intuition that the ca se desc ribed, thoug h it gives us a n instance of a justified, true be lief, does not g ive us a c ase of knowledg e. An oth e r g oo d th ou g ht e xpe ri me nt a pp e a rs in " Psy c ho log ism a nd B e ha vio ri sm, " in which B lock (1981) g ives a de scription of a ma chine that c an pass the T uring Test, but tha t is no t in te lli g e nt: Call a string of sentenc es whose me mbers ca n be ty ped by a human ty pist one afte r anothe r in an hour or less, a ty pa ble str ing of se nte nc e s. . . . [T ]his se t ha s a very larg e, but nonethele ss finite, number of me mbers. Consider the subse t of thi s se t w hic h c on ta ins a ll a nd on ly tho se str ing s w hic h a re na tur a lly int e rp re ta ble a s c on ve rs a tio ns in w hic h a t le a st o ne pa rt y ' s c on tr ibu tio n is se ns ibl e in t he se ns e descr ibed above . Call a string which ca n be under stood in thi s way a se ns ibl e string. . . . I ma g ine a se t of se ns ibl e str ing s r e c or de d o n ta pe a nd de plo y e d b y a ve ry sim ple machine a s follows. The interr og ator ty pes in sentenc e A . The mac hine sea rche s its li st of sensible string s, picking out those that beg in with A . I t then picks one of these A -i nit ia l st ri ng s a t r a nd om, a nd ty pe s o ut i ts s e c on d s e nte nc e , c a ll i t " B ." The interr og ator ty pes in sentenc e C . The mac hine sea rche s its li st, isolating the

PAGE 113

104 strings tha t start with A f o l l o w e d b y B f o l l o w e d b y C . I t picks one of these ABC initial st ring s and ty pes out its fourth sentenc e, and so on. . . . . . . [S ]uch a machine will have the c apac ity to emit a sensible seque nce of verba l outputs, whatever the ver bal inputs, and henc e it is intelli g ent ac cording to the ne oTu ri ng Te st c on c e pti on of int e lli g e nc e [(t he c a pa c ity to p ro du c e a se ns ibl e sequenc e of ve rbal re sponses to a seque nce of verba l stim uli, whatever they may be)]. (B lock 1981, 19-21) The re levant question that is implicit i n this scenar io is: "I s the machine intelligent? ," a nd the intuition t hat an idea l participant will likely have is, " No, the mac hine is not int e lli g e nt. " Th is i s in tui tiv e ly a g oo d th ou g ht e xpe ri me nt b e c a us e the re is e no ug h d e ta il about the situation that we c an have a cle ar intuition about what to say in response to the canonic al question. I t presents us with a c ountere x ample to the cla im that all something needs to be intelligent is an a ppropriate sequenc e of ve rbal output to a seque nce of verba l sti mul i. We c a n c on c lud e fr om t his ta rg e t sc e na ri o th a t so me thi ng mor e tha n ju st appropr iate ver bal output is necessa ry for intellig ence , and, so, being able to pass the Tu ri ng Te st i s n ot s uf fi c ie nt f or be ing int e lli g e nt. I f t his ta rg e t sc e na ri o w e re pr e se nte d in the way I have sta ted thoug ht experiments can be pr esente d, the re levant question that would be posed to the pa rticipant is: I s the machine descr ibed intellige nt? I t seems that we would sa y "no," and that we have a strong intuiti on that the answe r is "no." A third intuiti vely g ood thought e x periment is one that a ppear s in Grice (1961) wher e he w ants to arg ue that it is nece ssary for X 's p er ce i v i n g M tha t X ' s sense pe rce ption c a us a lly de pe nd on a sta te of a ff a ir s th a t in vo lve s M . The pa rticipant is g iven a descr iption of a targ et sce nario in which a clock' s being in an individual's f ield of vision sometimes does and sometimes does not ca use someone to pe rce ive the cloc k though the individual continuously has the visual experienc e of the clock, whic h is supposed to show tha t M must actually play a ca usal role in X ' s visual experience for X to actually perc eive

PAGE 114

105 M . T he n s he is g ive n th e fo llo wi ng ta rg e t sc e na ri o th a t sh ow s th a t a lth ou g h it is nece ssary for M to play a ca usal role in X ' s v isu a l e xpe ri e nc e , it is n ot s uf fi c ie nt: Suppose that it looks t o X as if ther e is a c lock on the shelf; wha t more is require d fo r i t to be tr ue to s a y tha t X sees a clock on the she lf? There must, one might say , actua lly be a c lock on the shelf w hich is in X ' s field of view , befor e X ' s ey es. But this does not seem to be e noug h. For it is l og ically conce ivable that ther e should be some method by which an e x pert c ould make it look to X as if ther e wer e a c lock on the shelf on oc casions whe n the shelf wa s empty : there mig ht be some appa ratus by which X ' s c or te x co uld be su ita bly sti mul a te d, or so me technique a nalog ous to post-hy pnotic sug g estion. (Gric e 1961, 69) A ca nonical question implicit in t his scenar io is, "I n this case, doe s the individual see the clock? " T he intuition that an ideal par ticipant would have, it see ms, is "No, the individual does not see the c lock." We can infe r fr om the participa nt's r esponse to the canonic al question that she thinks that for X to see the c lock, the cloc k must actually be play ing a role in produc ing the visual experience . This thought experiment shows that for the c ausal theor y of per ception to be true , that which the pe rson is having a visual experience of must actually be play ing a role in produc ing tha t experience . The thoug ht e xpe ri me nt c on tin ue s, ho we ve r, wi th a no the r t a rg e t sc e na ri o, wh ic h s ho ws tha t th is condition, although ne cessa ry , is not sufficient for se eing the clock. [A] s it stands it can ha rdly be suff icient. For in any particula r per ceptua l situation there will be objects other tha n that which would ordina rily be re g arde d as being perc eived, of which some state or mode of func tioning is ca usally rele vant to the occur renc e of a particula r sense -impression: this mi g ht be true of such objec ts as the per cipient' s ey es or the sun. So some re striction will have to be adde d to the analy sis of perc eiving which is under c onsideration. (Gric e 1961, 70) Although the way this case is desc ribed ther e is no rele vant question that might be asked, we c ould describe the ca se in a slig htly differ ent way , which will allow us to pose a question: Suppose someone is seate d in front of the c lock on the shelf, but doe s not have a direc t view of it. Ther e is a tele vision camera and a sc ree n in front of the c lock which produc e a holog raphic imag e of the clock on the she lf so that the individual

PAGE 115

106 has the same visual experience she would have had if ther e we re no c amer a and no scre en. A re levant question an e x perimente r mig ht ask is: "I s the clock' s causa l role in X 's perc eption of it sufficient f or X ' s p e rc e ivi ng the c loc k?" A n id e a l pa rt ic ipa nt, it s e e ms, would say "no, it is not enoug h. There are other thing s that play a ca usal role in X 's perc eption of the c lock." This second ha lf of the thoug ht experiment shows that the causa l role of the obje ct per ceive d is not the only thing that is ca usally rele vant to the pe rc e pti on of the ob je c t ( the c loc k is c a us ing the vis ua l e xpe ri e nc e bu t th e ind ivi du a l is no t pe rc e ivi ng the c loc k) , a nd so on e of the the se s o f t he c a us a l th e or y of pe rc e pti on is inadequa te. As a w hole, this thought experiment shows that althoug h one of the c on dit ion s o f t he c a us a l th e or y of pe rc e pti on is n e c e ssa ry fo r p e rc e ivi ng a n o bje c t, i t is not sufficient. This is a g ood thought e x periment be cause it includes all and only the rele vant details that ar e nee ded for a re levantly complete de scription of a ta rg et sce nario, a nd it i s c le a r w ha t c on c e pt i s in vo lve d in it. Now that we have se en some e x amples of g ood thought e x periments and ha ve g ott e n a lit tle c le a re r o n s ome ne c e ssa ry c on dit ion s f or a g oo d th ou g ht e xpe ri me nt, I wi ll lay out some nece ssary conditions for g ood thought e x periments. F irst, in a g ood thought experiment the participa nt is not forced to ha ve a c erta in intuit ion and is not biased in any way to be more inc lined to have one intuiti on rathe r than a nother. Tha t is, the e xpe ri me nte r s ho uld no t de sc ri be the c a se or a sk a qu e sti on in s uc h a wa y tha t it is obvious what the experimenter wants the pa rticipant to say in response to the ta rg et sc e na ri o. Se c on d, a g oo d th ou g ht e xpe ri me nt s ho uld inc lud e a ta rg e t sc e na ri o th a t is rele vantly completely and ca ref ully descr ibed. A re levantly complete de scription of a ta rg e t sc e na ri o, of c ou rs e , d oe s n ot r e qu ir e a c omp le te de sc ri pti on of the wo rl d, bu t on ly

PAGE 116

107 a complete descr iption of what is releva nt to the question that is asked. A ca ref ul descr iption of a targ et sce nario should not include irre levant or e x tra de tails that might confuse or sway a par ticipant' s intuit ions. Third, a g ood thought e x periment is clea r about which c oncept the pa rticipant should consult her intuitions about. Fourth, the question asked should be one that g ets at the rig ht intuit ions. Ag ain, I do not think that thi s is a c omp le te lis t of ne c e ssa ry c on dit ion s f or a g oo d th ou g ht e xpe ri me nt. Ea rl ie r i n th is c ha pte r, I ma de a dis tin c tio n b e tw e e n s tr on g c on c e ptu a l a na ly sis A and we ak conc eptual ana ly sis. W here wea k conce ptual analy sis y ields nece ssary and A sufficie nt application conditions, wea k conce ptual analy sis only trac es conne ctions betwee n conce pts. I n the pre vious section, I listed four thing s that we c an use thoug ht e x p e r i m e n t s t o d o : ( a ) t o s h o w t h a t a p r o p o s e d n e c e s s a r y c o n d i t i o n i s n o t n e c e s s a r y; (b ) t o s ho w t ha t a pr op os e d s uf fi c ie nt c on dit ion is n ot s uf fi c ie nt; (c ) t o s ho w t ha t a proposed suff icient condition is sufficient; and ( d) to show that a pr oposed nec essar y R condition is necessa ry . We arrive at a we ak conc eptual ana ly sis by using thoug ht R experiments to do any one of ( a)–( d), and w e ar rive a t a strong conce ptual analy sis by using thoug ht experiments to do all of (a)–( d), sometimes for more than one pr oposed condition. We u se tho ug ht e xpe ri me nts to d o o ne or a ll o f ( a )– (d ). Do ing a ny of (a )– (d ) i s clea rly doing c onceptua l analy sis. W hen we imag ine targ et sce narios and r evea l our intuiti ons about them or cr eate targ et sce narios for others to imag ine, nothing a bout our intuiti ons or our a bility to cre ate tar g et sce narios depe nds on any particula r experienc e that we may have. T hus, the knowledg e that we acquir e fr om conducting thought experiments is conceptua l knowledg e.

PAGE 117

108 I n the next chapter, I will go into more de tail about the fra mework f or thinking about thoug ht experiments that I have c rea ted. I will offer an ac count of c oncept po sse ssi on , w hic h p la y s a n im po rt a nt r ole in w ho I thi nk sh ou ld b e pa rt ic ipa nts in thought e x periments. I also introduce the notion of proper perf ormanc e of a thought e xpe ri me nt a nd a dd re ss s ke pti c ism a bo ut d oin g tho ug ht e xpe ri me nts .

PAGE 118

109 CH APT ER 3 DETAI L S OF THE F RAMEWORK: C ONCEPT POSS ESSI ON AND SOURCES OF ERROR The purpose of this chapter is to fill i n some of the de tails of the fra mework f or defe nding a traditional approa ch to thinking a bout thought experiments. When thinking a bo ut h ow we sh ou ld d o th ou g ht e xpe ri me nts we sh ou ld g e t c le a r o n w ho sh ou ld participa te in thought e x periments if we want their r esults to be useful. Chara cter izi ng the notion of conc ept possession will be helpful in doing that. We should also want an a c c ou nt o f h ow we c a n a vo id f a lli ng int o e rr or in d oin g tho ug ht e xpe ri me nts . Sui table Candid ates and Concept P ossession I n the pre vious chapter , I talked a lot about wa y s participants may answe r var ious c a no nic a l qu e sti on s, bu t I ha ve y e t to g ive c on dit ion s u nd e r w hic h s ome on e ha s r e lia ble i n t u i t i o n s ( i . e . , i n t u i t i o n s f r o m w h i c h w e c a n r e l i a b l y d r a w p h i l o s o p h i c a l c o n c l u s i o n s ) . In this section, I will give a nd motivate the conditions under whic h an individual is an appropr iate subjec t of, or a pa rticipant in, a thoug ht experiment, in the sense that we would be justified in acc epting the re sults from the experiment as conc eptual knowledg e. I n d oin g thi s, I wi ll a rt ic ula te wh a t so rt of min ima l kn ow le dg e is n e e de d f or so me on e to be inv olv e d in a tho ug ht e xpe ri me nt. Concept P ossession Who wo uld be a n id e a l, o r a t le a st a su ita ble , p a rt ic ipa nt i n a tho ug ht e xpe ri me nt, (that is, how should the experimenter c hoose his subjects? ). Since wha t we do in thoug ht experiments is try to ge t at individuals' intuitions a bo ut w he the r a c on c e pt a pp lie s o r f a ils

PAGE 119

110 to a pp ly , o r w he the r i t is ind e te rm ina te tha t it do e s, the e xpe ri me nte r s ho uld c ho os e someone who c an ac tually have r eliable intuitions about the conce pt in question. An intuiti on, in particular , a sc e na rio intuiti on, as I stated ea rlier, is an unme diated cog nitive seeming about a tar g et sce nario. Sce nario intuitions are not made on the basis of philosophical theorizing, and the y are not kneejerk or g ut rea ctions to a targ et sce nario. Rather, one should care fully consider w hether the conc ept applies or f ails to apply , or whether it is i ndeter minate, but not allow any theory about the conc ept' s application conditions to causally aff ect one ' s considera tion. I f one ha s a ce rtain amount of philosophical backg round, she must not allow it to affe ct the wa y she re sponds to a canonic al question in a thoug ht experiment. Be ing a ble to car efully consider the se matters r equire s a ce rtain deg ree of possession of the conc ept in question, so philosophers must be car eful a bout who they ask for intuiti ons if they want thoug ht experiments to be useful in acquir ing c onceptua l knowledg e. I n this section, I will lay out two levels of conc ept possession that are re le va nt t o w ho we sh ou ld c ho os e to b e pa rt ic ipa nts in thought experiments. Ther e ar e all sorts of things tha t make someone an unsuitable pa rticipant in a thoug ht experiment. I do not mean to list all of these in my acc ount here , but just t o g ive an idea of the kinds of thi ng s th a t w ou ld m a ke so me on e a n u ns uit a ble pa rt ic ipa nt. There are a fe w things w e nee d to ge t clear on when de termining what deg ree of c on c e pt p os se ssi on is r e qu ir e d f or on e to h a ve re lia ble int uit ion s w he n p a rt ic ipa tin g in tho ug ht e xpe ri me nts , a nd wh o h a s th a t a nte c e de nt k no wl e dg e . F ir st, the pa rt ic ipa nt m us t b e a c o mp e te n t s p e a k e r o f th e la n g u a g e th e c o n c e p t i n q u e s ti o n is e xp r e s s e d in . We would not ask someone to ente rtain a thoug ht experiment in English, which uses a n

PAGE 120

111 Eng lish word to express the conce pt in question, if the individual is not a compete nt sp e a ke r o f E ng lis h. B e ing a c omp e te nt s pe a ke r o f t he la ng ua g e re qu ir e s a ba sic knowledg e of g rammar and voca bulary so that he or she can utter and under stand intelligible sente nces. B ut, this i s not enoug h. I n addition, we should want the p ar t i ci p an t t o b e a b l e t o co n v er s e f l u en t l y wi t h o t h er co m p et en t s p ea k er s o f t h e l an gu age using the word that e x presse s the conc ept in question. Even be ing a competent spe aker who ca n fluently conver se with other c ompetent speake rs is not enoug h for someone to be an idea l, or at lea st a suitable, ca ndidate for participa tion in a thought experiment. The pa rticipant must also po sse ss the conc ept that the thoug ht experiment is try ing to g et clea r about, whe re a conce pt is the sense of a p r e d ic a te . I t i s n o t c le a r th a t c o mp e te n c e , a s c h a r a c te r iz e d a b o v e , s u f f ic e s f o r th is . We mus t, t he n, a t th is p oin t la y ou t th e c on dit ion s u nd e r w hic h a n in div idu a l po sse sse s a conce pt that are rele vant to participa tion in a thought experiment. I have a lrea dy briefly addre ssed the issue of c oncept possession in discussions of Be aler ' s and Peac ocke' s theories of c oncepts. B eale r admits a distinction betwee n two levels of c oncept possession, nominal possession and full possession, wher e the f ormer require s that the individual have propositional attitudes towar d propositions that have the conce pt as part of its content and allows for some misunderstanding of the c oncept, a nd the latter r equire s that there is no misunderstanding or incomplete unde rstanding of the conce pt. Peacoc ke, on the other hand, g ives an a ccount of conce pt possession accor ding to which an individual possesses a conce pt if and only if she finds ce rtain infer ence s primitively compelling .

PAGE 121

112 I borrow this terminolog y from B eale r, but I put it to di ffe rent use 1 (B eale r 1998a). Whe n I us e the te m ' kn ow -h ow ' , I a ssu me the tr a dit ion a l Ry le a n a na ly sis 2 (R y le 19 46 , 1 98 4) . I f R y le ' s a na ly sis is u ns a tis fa c tor y , I wo uld g ive up ' kn ow -h ow ' or us e it i n a te c hn ic a l se ns e bu t it wo uld no t a ff e c t my a rg ume nt; it w ou ld j us t a ff e c t ho w I e xpre ss it (Stanley and Willi amson 2001). I want to ar g ue that ther e ar e at lea st two differe nt levels of conc ept possession, na me ly , me re no min a l po sse ssi on a nd fu ll p os se ssi on tha t w e sh ou ld p a y a tte nti on to 1 when c onsidering who would be a n ideal, or a t least a suitable, pa rticipant in a thoug ht experiment. The conditions for c oncept possession that I g ive ar e g iven as c apac ities, or know-how. I do this because I think that the more intere sting re quirements for 2 pa rt ic ipa tio n in tho ug ht e xpe ri me nts inv olv e c e rt a in a bil iti e s. The fir st level of conc ept possession that I want to discuss is mere nominal conce pt possession. The kind of nominal possession I have in mind is the kind of c on c e pt p os se ssi on tha t B ur g e (1 97 9) dis c us se s in his a rt hr iti s c a se : A g ive n p e rs on ha s a la rg e nu mbe r o f a tti tud e s c omm on ly a ttr ibu te d w ith c on te ntclause s containing "a rthritis" in oblique occ urre nce. F or example, he thinks (cor rec tly ) that he ha s had ar thritis for y ear s, that his arthritis in his wrists and fing ers is more pa inful than his arthritis in his ankles, that it is better to have arthritis than ca ncer of the liver, tha t stiffening of the joints is a sy mptom of arthritis. . . . I n addition to these unsurprising attitudes, he thinks falsely that he has deve loped ar thritis in t he thig h. (B urg e 1979, 77) I n this case, if the re is a se nse in which the pe rson possesses the c oncept of ar thr iti s , he clea rly does not fully possess it, though ther e ar e cir cumstance s in which he c an use the t er m t h at ex p re s s es t h e c o n ce p t co rr ec t l y . I n o rd i n ar y co n v er s at i o n i n v o l v i n g 'ar t h ri t i s ', fo r e xamp le , th e pe rs on in t he B ur g e c a se c a n, fo r t he mos t pa rt , ta lk a bo ut a rt hr iti s wi tho ut g ro ss m ish a p ( i.e ., he kn ow s e no ug h a bo ut ar thr iti s to avoid obvious conce ptual

PAGE 122

113 Re le va ntl y c omp le te ly de sc ri be d ta rg e t sc e na ri os c a n b e e ith e r a c tua l or me re ly 3 possible. One mig ht object to my char acte rization of nominal concept possession on the 4 g ro un ds tha t I le a ve no ro om f or no nlin g uis tic a nim a ls t o p os se ss c on c e pts . I n th is dis se rt a tio n, I a m in te re ste d in a no tio n o f f ull c on c e pt p os se ssi on re le va nt t o li ng uis tic ag ents which he lps ill uminate their pra ctice of using thoug ht experiments to ex plore modal questions. The issue of c oncept possession by non-ling uistic animals is something that should be addre ssed in some other c ontext . absurdities). I char acte rize this kind of concept possession as mer e nominal conc ept po sse ssi on : A m e re ly no mi na lly po sse sse s c on c e pt C iff (a ) there is some word, w , which e xpre sse s C in a lang uag e L , and A can use w in ordinary conver sation and pass a s a compe tent speake r of L wi t h re gar d s t o u s i n g w ; a nd (b ) t he re a re re le va ntl y c omp le te ly de sc ri be d ta rg e t sc e na ri os su c h th a t A ne e ds to c on su lt a lin g uis tic 3 authority in order to a rrive a t a justified judgme nt as to whether C a pp lie s, fa ils to apply , or it is objectively indeterminate or epistemica lly indeterminate whether C a pp lie s. 4 Th a t is to s a y , th e re a re sc e na ri os tha t A cannot a nswer r eleva nt questions about on the basis of his competenc e but for w hich there are deter minate answe rs g iven the conc ept. I will ex plain what I mean by ' objectively indeterminate ' and ' epistemically indeterminate ' be low . We c a n h a ve a te st t ha t w ou ld s ho w w he the r s ome on e me re ly no min a lly po sse sse s a c on c e pt. Th is i s e a sie r t o e xpla in i f w e re la te me re no min a l po sse ssi on to partially understanding a wor d. For example, consider the w ord ' brisket' , another of B ur g e ' s e xamp le s. An ind ivi du a l ma y kn ow ma ny thi ng s a bo ut b ri sk e t, n a me ly , th a t it comes fr om (among others) c ows, that it is red, and so on. B ut, the individual may not kn ow wh a t pa rt of the c ow the pie c e of me a t c ome s f ro m. T hu s, the re a re c e rt a in c a se s in which the individual will not know whether the term ' brisket' applies (e .g ., a ca se whe re the ind ivi du a l is tol d th a t th e re is a pie c e of me a t f ro m th e br e a st o f a n a nim a l th a t is cooked in suc h and such a way and then a sked whe ther that piec e of me at is brisket).

PAGE 123

114 A r e le va ntl y c omp le te ly de sc ri be d ta rg e t sc e na ri o c a n b e e ith e r a c tua l or me re ly 5 possible. The wa y I have de fined mer e nominal possession and full possession, when one 6 fully possesses a c oncept, she does not mere ly nominally possess it. That is, full concept possession does not amount to merely nominally possessing it plus some additional a bil iti e s. One wa y of re ading clause (b) may make c lause (a ) see m redunda nt. Consider 7 the following : conce ptually true propositions are true in all possible worlds. I f one possesses a c oncept, then one alre ady has the a bility (by and lar g e, at lea st) to determine when the c oncept a pplies or not, or whe ther it is indeterminate w hether the conc ept a pp lie s to a g ive n ta rg e t sc e na ri o. Th a t se e ms r ig ht. I t ma y loo k a s if the a bil ity to evalua te targ et sce narios and de termine whe ther a conce pt applies, etc. is the g round of the pe rs on ' s r e c og nit ion of sim ple c on c e ptu a l tr uth s. I f t ha t is so , th e n it loo ks a s if requiring that one be a ble to rec og nize sim ple conc eptual truths is redunda nt beca use by b e in g a b le to c o r r e c tl y e v a lu a te ta r g e t s c e n a r io s , o n e is th e r e b y a b le to r e c o g n iz e This is a question that someone who fully understands the word would be able to answ er but someone who only partially understand the word may not be able to a nswer. When an individual mere ly nominally possesses a c oncept, she will not be able to answe r all of the re le va nt q ue sti on s a bo ut w he the r t he c on c e pt a pp lie s o r f a ils to a pp ly , o r w he the r i t is ind e te rm ina te tha t it do e s. I n a dd iti on , s he wi ll l ike ly no t r e c og nize su ff ic ie ntl y sim ple conce ptually true propositions involving the c oncept that we would expect someone who fully possesses the c oncept to be a ble to. We want someone to pa rticipate in a thought experiment involving ce rtain conc epts only if she possesses the conce pts more than mere ly nominally . The sec ond level of c oncept possession that I want to discuss is full conce pt possession: A fu lly po sse sse s c on c e pt C if and only if (a) A is able to rec og nize a priori as true s u ff i ci en t l y s i m p l e c o n ce p t u al l y t ru e p ro p o s i t i o n s i n v o l v i n g C a nd oth e r c on c e pts which A fully possesses; and (b) A can a ccur ately (and w ith just ification) sa y about any rele vantly completely descr ibed targ et sce nario whether C a pp lie s, 5 fa ils to a pp ly , o r i t is ob je c tiv e ly ind e te rm ina te or e pis te mic a lly ind e te rm ina te whether C a pp lie s. 6, 7

PAGE 124

115 conce ptual truths. Howeve r, it may be that the a bility to evaluate targ et sce narios is not sufficie nt for having the ability to rec og nize sim ple conc eptual truths. I n fac t, it may be the other w ay around—the ability to rec og nize sim ple conc eptual truths may be re quired to e va lua te ta rg e t sc e na ri os . I n o rd e r t o k e e p th is p os sib ili ty op e n, we sh ou ld m a ke bo th (a) and (b) nece ssary for f ull concept possession. Actually , (a) has the same modal element. A n ee d s t o b e a b l e t o (e v en t h o u gh 8 s h e m ay n o t ) r ec o gn i z e c o n ce p t u al l y t ru e p ro p o s i t i o n s i n v o l v i n g C to possess it. For example, A may not rec og nize sufficiently simple conce ptually true propositions i n v o l v i n g C if she is not pay ing a ttention. Th e fi rs t c la us e se e ms s tr a ig htf or wa rd e no ug h, bu t th e se c on d c la us e ne e ds so me clar ification. Ther e is quite a bit pac ked into (b). F irst, there is the modal e lement of (b) , na me ly tha t A can a c c ur a te ly sa y wh e the r a c on c e pt a pp lie s, fa ils to a pp ly or it i s objectively indeterminate or epistemica lly indeterminate whether C applies. Second, there are the notions of objective indete rminacy and epistemic inde terminac y . Third, the re is t he c la im t ha t A must be able to say whether C applies, etc . to any re le va ntl y completely descr ibed targ et sce nario. There are two functions that the modal ele ment serve s in the above de finition of fu ll c on c e pt p os se ssi on . F ir st, it a llo ws on e to m a ke a mis ta ke a bo ut C 's application, and, sec ond, it allows us to give c onditions under which A sh ou ld b e a ble to 8 judge acc urate ly whether C applies. A ma y ma ke a mis ta ke a bo ut C ' s a pp lic a tio n if sh e is not pay ing c are ful attention to the descr iption of the targ et sce nario or if she depe nds on the a c tua l w or ld t o d e te rm ine C' s in te ns ion . Si nc e we a re fa lli ble , a nd I de fi ne fu ll c on c e pt p os se ssi on in t e rm s o f a n a bil ity to d e te rm ine wh e the r a c on c e pt a pp lie s, fa ils to a pp ly , o r i t is ob je c tiv e ly ind e te rm ina te or e pis te mic a lly ind e te rm ina te wh e the r i t do e s, we should make room for e rror .

PAGE 125

116 A has a g rea ter a bility to work thing s out if she g ets the wrong answe r at fir st, but 9 upon ref lection ca n see w here she we nt wrong and g et the rig ht answer . See 10 Be aler 2002b for a discussion of this abilit y to work thing s out. L ater , I ref er to those individuals who ca n only say whether a conc ept applies 11 wh e n a sk e d u nd e r s uit a ble c on dit ion s ' su ita ble c a nd ida te s. ' I do thi nk tha t a su ita ble c a nd ida te is o ne wh o w e c a n r e ly on to g ive us int uit ion s a bo ut t a rg e t sc e na ri os . The modal e lement in (b) a lso allows us to say something a bout the conditions under w hich A sh ou ld b e a ble to m a ke c or re c t ju dg me nts a bo ut C . A c a n s a y c or re c tly whether C applies, etc ., rela tive to a suitable applica tion of resourc es, for example, additional conce pts, a g rea ter a bility (than the a vera g e compe tent speake r of the lang uag e) to wor k things out, and being in an epistemic position make judg ments about 9 C ' s a pp lic a tio n. Th e re is m or e wo rk tha t c a n b e do ne in c ha ra c te ri zing wh a t th e a bil ity 10 itself is. For e x ample, whe n we sa y that someone can do these thing s, we c an think about exactly what this impli es. I n addition, we ca n make pr og ress by char acte rizing situations under w hich the per son wou ld make c orre ct judg ments. We m a y re a d ( b) a s me a nin g tha t w he n s he is a sk e d u nd e r s uit a ble c on dit ion s A can sa y whether C a pp lie s, e tc . On e mig ht a rg ue (a s I wi ll) tha t so me tim e s th os e 11 su ita ble c on dit ion s a nd be ing the ri g ht e pis te mic po sit ion inc lud e be ing pr ov ide d w ith rele vant information about the a ctual wor ld. Now, surely someone who is an ideal c a nd ida te fo r p a rt ic ipa tio n ( I wi ll e xpla in w ha t a n id e a l c a nd ida te a s o pp os e d to a su ita ble c a nd ida te is b e low ) i n a tho ug ht e xpe ri me nt w ill no t ne e d a ny e mpi ri c a l in fo rm a tio n to g ive the cor rec t answer to a ca nonical question. B ut, there is no re ason to think that someone who ne eds such infor mation is not a su ita ble candida te for pa rticipation in a thought e x periment (tha t is, someone who, with cer tain rele vant empirica l information (a nd c e rt a in a bil iti e s) , c a n g ive re lia ble a ns we rs to c a no nic a l qu e sti on s) . Wh e n th e re is

PAGE 126

117 information about the a ctual wor ld that may be re levant to eva luating a targ et sce nario, the experimenter ma y prompt the participa nt with what I will call ' stipulated backg round information' . Stipul ated ba ckg round informa tion is i nformation about the a ctual wor ld (whethe r acc urate or mere ly supposed) that may help a pa rticipant in a thoug ht experiment answer a ca nonical question. Not eve ry thought e x periment include s (or should include) a step in which the experimenter asks the par ticipant to suppose something a bout the actua l world. Althoug h backg round informa tion is i n many case s irrele vant to a thoug ht experiment, when it is, in cases of natura l kind terms, which express natura l kind c on c e pts , it c a n h e lp a pa rt ic ipa nt t o g ive a n a ns we r t o a c a no nic a l qu e sti on . I n th os e case s, the experimenter should ask the participa nt to make a supposition about the ac tual world. F ur the rm or e , I sa y tha t st ipu la te d b a c kg ro un d in fo rm a tio n is inf or ma tio n th a t may be usef ul in evaluating a tar g et sce nario be cause surely one could stipulate that in the a c tua l w or ld s uc h a nd su c h c on dit ion s o bta in, e ve n w he n th e ob ta ini ng of tho se c on dit ion s is no t su ff ic ie nt t o d e te rm ine the int e ns ion (a fu nc tio n f ro m w or lds to extensions, or truth values) of the c oncept. F or example, with the conc ept of water , we could stipulate that there are no dog s in the actua l world, but that would clea rly be irrele vant to whether the conc ept of water applied to any thing. Althoug h stipulated backg round informa tion may be helpf ul in evaluating the targ et sce nario, it nee d not be part of the targ et sce nario. Giving a par ticipant rele vant stipulated bac kg round inf or ma tio n s ho uld he lp m ini mize e pis te mic ind e te rm ina c y , s inc e wh a t ma ke s it epistemically indeterminate whether a conc ept applies is a lac k of informa tion about the actua l world.

PAGE 127

118 I stated above that sometimes we run thoug ht experiments on words or 12 predic ates, whic h are sometimes vag ue. I t he lps in t a lki ng a bo ut t he uti lit y of tho ug ht e xpe ri me nts to i ntr od uc e c on c re te 13 proce dures f or distinguishing the " I don' t know" a nswers of the epistemic indete rminacy so rt to c a no nic a l qu e sti on s in to s e ma nti c ind e te rm ina c y a nd pu re ly e pis te mic ind e te rm ina c y . I n c a se s w he re the a ns we r t o a c a no nic a l qu e sti on is s e ma nti c a lly indeterminate , no more de tails in the targ et sce nario a nd no amount of stipulated backg round informa tion is going to make a dif fer ence to the "I don' t know" r esponse. Th e a ns we rs to c a no nic a l qu e sti on s th a t a re se ma nti c a lly ind e te rm ina te a re g oin g to b e in t h o u g h t e x p e r i m e n t s i n v o l v i n g v a g u e t e r m s . In c a s e s o f p u r e l y e p i s t e m i c i n d e t e r m i n a c y, adding further details may help a pa rticipant answe r a c anonica l question. W e ca n dis tin g uis h th e se tw o k ind s o f i nd e te rm ina c y by do ing mor e tho ug ht e xpe ri me nts wi th the extra details that would, if the a nswer is pure ly epistemically indeterminate , help the participa nt give a def initive answer to the c anonica l question. B e fo re we c on tin ue ou r d isc us sio n a bo ut t he ro le of ' c a n' in c la us e (b ) o f m y definition of full conc ept possession, a bit of cla rifica tion is needed to explain the two differ ent way s that it can be inde terminate w hether a conc ept applies. F or some thoug ht experiments on words, even w hen a ta rg et sce nario is re levantly completely descr ibed, 12 it is open whether or not it applies (i.e., it is objectively indeterminate ). This happe ns wh e n a ta rg e t sc e na ri o d e sc ri be s a tr uly bo rd e rl ine c a se or wh e n th e wo rd or pr e dic a te in question is vag ue. Epistemic indeter minacy can be understood as the following : give n wh a t A knows about the a ctual wor ld, it is not determined w hether C applies. I t can be epistemically indeterminate whether a conc ept applies, for example, if there is something a bo ut t he wo rl d th a t A do e s n ot k no w, bu t w hic h c a n a ff e c t C ' s a pp lic a tio n. Th is w ill 13 most li kely happen in thoug ht experiments involvi ng natura l kind concepts. Althoug h A need not ha ve the r eleva nt information about the wor ld to determine whe ther the c oncept applies, A needs to be sensitive to the possibili ty (if ther e is a possibility ) that the wa y things a re in the a ctual wor ld help deter mine whether C a pp lie s, fa ils to a pp ly , o r i t is epistemically indeterminate whether C applies. A has to be sensitive in the following

PAGE 128

119 One mig ht expect 'e pistemic indeterminac y ' to mean that the de scription of the 14 scena rio does not include e noug h information to say whether C a pp lie s o r n ot. So i t is epistemically indeterminate . But this is ruled out by my alre ady stipulating that the descr iption is rele vantly complete. I f someone sa y s "I don' t know" it could be an a p p r o p r i a t e a n s w e r f o r a t l e a s t t h r e e d i f f e r e n t r e a s o n s : o b j e c t i v e i n d e t e r m i n a c y, incomplete desc ription, or a nee d for infor mation about the ac tual world. way : if prompted appr opriately with various stipulated suppositions about the chara cter of the a ctual wor ld, then A would rea liz e the ne ed to judg e " epistemically indeterminate " fo r c e rt a in t ho ug ht e xpe ri me nts . 14 We might, then, cla rify clause (b) of the def inition of full concept possession as fo llo ws : (b' ) A c a n a c c ur a te ly (a nd wi th j us tif ic a tio n) sa y a bo ut a ny re le va ntl y c omp le te ly descr ibed targ et sce nario whe ther C a pp lie s, fa ils to a pp ly , o r i t is ob je c tiv e ly indeterminate whether C applies; and whe n g iven such a nd such stipulated backg round informa tion, A can a ccur ately (and w ith just ification) sa y about any re le va ntl y c omp le te ly de sc ri be d ta rg e t sc e na ri o th a t it is e pis te mic a lly indeterminate whether C a pp lie s w he n th e sti pu la te d c on dit ion s a re no t e no ug h to fix the intension. B ut, thi s se e ms t oo de ma nd ing , s inc e on e mig ht f a il t o r e c og nize tha t th e ri g ht a ns we r t o the ca nonical question is "e pistemically indeterminate " f or re asons that do not have a n y t h i n g t o d o w i t h a f a i l u r e t o p o s s e s s t h e c o n c e p t , b u t w i t h o n e ' s e p i s t e m i c p o s i t i o n . In t h e f o l l o wi n g, I p re s en t t wo s i t u at i o n s i n wh i ch o n e's ep i s t em i c p o s i t i o n af fe ct s o n e's ability to make a c orre ct judg ement about a targ et sce nario. Consider the following situation: Sup po se the c on c e pt C is such that if P is true in the actua l world, C is assigne d function F , and if P* is true in the ac tual world, C is assigne d function F *. I should point out that I am assuming a coa rseg raine d approa ch to conc ept individuation, which I explained in the previous cha pter in the sec tion entitled, "I ndividuating Concepts." The ba sic idea be hind a coa rseg raine d approa ch is that one c oncept c an have tw o d if fe re nt i nte ns ion s.

PAGE 129

120 This is a conseque nce of appea ling to a c oarse -g raine d approa ch to the 15 ind ivi du a tio n o f c on c e pts . Sup po se fu rt he r t ha t a n a g e nt S is such that both of the following are true of he r: (1) I f S wer e told to suppose that P is t he ca se, S would make judg ments that confor m to the assig nment of func tion F to t he c on c e pt. (2) I f S wer e told to suppose that P* is t he ca se, S would make judg ments that confor m to the assignme nt of function F* to t he c on c e pt. I n such a c ase, w hether the conc ept applies, e tc. depe nds on what thing s are like in the a c tua l w or ld. I f w e a c c e pt ( b' ) a nd su pp os e tha t S is not given a ny stipulated backg round information, then if it is indeterminate w hat func tion should be assigne d to the conce pt g ive n w ha t S kn ow s ( no t w ha t sh e is t o s up po se ), we wo uld e xpe c t S to a ns we r, " I t is e pis te mic a lly ind e te rm ina te ." B ut i t do e s n ot l oo k a s if we sh ou ld e xpe c t S to a ns we r i n this way beca use re lative to the bac kg round informa tion she is given, she makes the c or re c t ju dg me nt. To de te rm ine tha t it is S ' s epistemic position that runs interfer ence with her pe rfor mance of the thoug ht experiment we have to see how S re a c ts n ot j us t to one bit of stipulated bac kg round informa tion, but to contrasting pa irs of such. This is not pr ob le ma tic , o f c ou rs e , s inc e we of te n n e e d to pe rf or m se ve ra l th ou g ht e xpe ri me nts in order to determine w hether a par ticipant possesses the c oncept in question. Al te rn a tiv e ly , w e mig ht s a y tha t w e sh ou ld n ot t hin k th a t S should have a nswere d "e pistemically indeterminate " be cause once some one is prompted with stipulated ba c kg ro un d in fo rm a tio n w e sh ou ld c ou nt t he inf or ma tio n a s a bit of kn ow le dg e . T his kind of example helps illust rate the kind of conc eptual knowledg e we arr ive at whe n we evalua te targ et sce narios: I f the a ctual conte x t is such and such, then, if the targ et sc e na ri o is su c h a nd su c h, the n th e c on c e pt a pp lie s, fa ils to a pp ly , o r i t is ind e te rm ina te wh e the r i t a pp lie s. 15

PAGE 130

121 (b' ) wa s just a way of cla rify ing w hat (b) a lrea dy say s. We simpl y do not have 16 to mention all of the complications that come w ith (b) all of the time. The following situation is another in which S may not be able to make corr ect judgme nts about the applica tion of C beca use of he r epistemic position. L et P** be some proposition compatible with both P and P*. Now say we a sk S t o s t i p u l a t e t h a t P * * i s t r u e i n t h e a c t u a l w o r l d a n d c o n s i d e r a t a r g e t s c e n a r i o . It ma y be tha t S na tur a lly a ssu me s th a t P is the c a se , s inc e P** is c omp a tib le wi th P , and, say , S alre ady thinks P is in fact the c ase. I n this sit uation, S is li kel y to mak e ju dgmen ts in con for mi ty wi th th e ass ignm ent of F; that is, S inacc urate ly say s whether the conc ept applies, but she is justified in doing so. She wo uld no t be a uto ma tic a lly a stu te e no ug h to sa y , " Oh , it ' s e pis te mic a lly indeterminate beca use g iven what y ou' ve said, it might y et be that P* is true in the ac tual world!" I n other wor ds, she might not be se nsitive to the epistemic indeterminac y option if she is, in eff ect, c ompatibly with what she is told to ex plicitly suppose, re ly ing on backg round assumptions about what the a ctual wor ld is li ke. Someone might a rg ue that eve n if y ou include a supposition about the ac tual world, no such e x plicit sti pulation in practice is so complete as to rule out the inf luence , on the ag ent' s judgme nts, of other a ssumed backg round knowledg e, the influe nce of wh ic h s he ma y be qu ite un a wa re . I n s uc h s itu a tio ns , w e sh ou ld a sk the pa rt ic ipa nt t o consider dif fer ent (but re lated) thoug ht experiments, for example, ones that have the same c anonica l question and targ et sce nario but have differ ent stipulated bac kg round information, and se e how a nswers c hang e, with the hope tha t we ca n bring forth any ba c kg ro un d a ssu mpt ion s th e pa rt ic ipa nt i s g ro un din g he r a ns we rs in. To a vo id t he se kinds of complications, we ma y need to sa y that if being prompted by a stipulation about the ac tual world helps us to have reliable intuiti ons, then we must consider various rela ted thoug ht experiments (that is, thought experiments with differ ent stipulated backg round informa tion), and ac cept the or igina l formulation of (b) . 16

PAGE 131

122 Ar g ua bly , w e mig ht s a y tha t if we re qu ir e d th ou g ht e xpe ri me nts in g e ne ra l to 17 g ive us direc t answer s to questions about our conce pts, then we would be requiring too much of thoug ht experiments. We might do this in a situation where the way the ac tual world is can a ffe ct the 18 wa y on e a ns we rs the c a no nic a l qu e sti on bu t w he re we do no t kn ow the wa y the wo rl d a c tua lly is. Al tho ug h s tip ula te d b a c kg ro un d in fo rm a tio n ma y he lp m ini mize e pis te mic indeterminac y , it will not alway s lead to re sults that give us a nswers to que stions about whether the targ et sce nario pre sents us with a situation that shows the corre ct applica tion of the c oncept. I f, for example, the experimenter tells the pa rticipant to su pp os e that the 17 a c t u a l w o r l d i s a c e r t a i n w a y, b u t i t i s n o t c l e a r t h a t t h e a c t u a l w o r l d r e a l l y i s t h a t w a y, 18 the answe r that the pa rticipant g ives to the ca nonical question likely will not tell us any thing dire ctly about the applica tion conditions for the conc ept deter mined by the a c tua l c on te xt. Why ? B e c a us e the re is n o te lli ng wh e the r w ha t w e we re a sk e d to suppose is true about the a ctual wor ld. Supposi ng that the ac tual world is a ce rtain wa y in case s where we do not know wha t the actua l world is like in that respec t may help lend plausibilit y to cer tain positions we otherwise mig ht think thought experiments rule out, as I will ex plain in my analy sis of the zombi e thoug ht experiment. I f informa tion about the actua l world ca n aff ect the r esults of thoug ht experiments, as I will arg ue it does, then we ne e d to ta ke a c los e r l oo k a t c e rt a in t a rg e t sc e na ri os . At this point , I should take a step ba ck and ma ke cle ar the rela tion and the c on tr a st b e tw e e n me re no min a l a nd fu ll c on c e pt p os se ssi on . F ir st, le t us loo k a t c la us e (a) of ea ch of the definitions. The ability to rec og nize sufficiently simple conce ptual truths is parallel to the ability to use the re levant wor d in ordinary conver sations without g ross mishap. I n the linguistic c ase, tha t is, mere nominal conc ept possession, the per son

PAGE 132

123 kn ow s e no ug h to a vo id o bv iou s c on c e ptu a l a bs ur dit ie s. I n th e no nli ng uis tic c a se , th a t is , full conce pt possession, the person re cog nizes sim ple conc eptual truths. The r eason the individual in Burg e' s arthritis case does not fully possess the conc ept of ar thr iti s is that he has a fundame ntal misunderstanding about ar thritis that would prevent him from being ab l e t o re co gn i z e a l l s u ff i ci en t l y s i m p l e c o n ce p t u al l y t ru e p ro p o s i t i o n s i n v o l v i n g C and other c oncepts that the individual fully possesses: minimally , he does not mee t condition (a) . The main diff ere nce be tween me re nominal a nd full conce pt possession is t hat so me on e wh o f ull y po sse sse s a c on c e pt c a n s a y wh e the r i t a pp lie s, fa ils to a pp ly , o r i t is objectively indeterminate or epistemica lly indeterminate whether it applies, simply by ref lecting on the targ et sce nario (w ith, perhaps, some stipulated ba ckg round informa tion). We might say that to fully possess a conc ept, one must be fully linguistically competent with the term that expresses it. To be f ully linguistically competent with a te rm that expresses a c oncept, ther e should not be any misunderstanding s or lack of understanding of the wor d. On the other hand, someone who mere ly nominally possesses a c oncept ha s to resort to a ling uistic authority to answer a ca nonical question about a targ et sce nario. F ur the rm or e , s ome on e wi th m e re no min a l po sse ssi on ma y be a ble to a ns we r a ll t he se questions with information about the ac tual world, too, but only beca use the infor mation about the ac tual world inc lud e s sp e c ifi c all y semantic infor mation, such as " the word means such a nd such." The individual who fully possesses the c oncept may need information about the a ctual wor ld, but never spe cifica lly semantic infor mation. L et us consider an example of a n individual who has full possession of the conce pt of knowledge : for any targ et sce nario involving the conc ept of knowledge , the ind ivi du a l w ill , u nd e r o pti ma l c on dit ion s ( I sp e ll t his ou t in the ne xt se c tio n, in m y

PAGE 133

124 discussion about ideal and suitable c andidates) , have r eliable, thoug h not infallible, intuiti ons about whethe r the c oncept of knowledge a pp lie s, fa ils to a pp ly , o r i t is indeterminate whether it applies. I n the Smith and J ones targ et sce nario, the individual will likely judge that Smit h does not know that the man w ho will ge t the job has ten coins in his pocket (Ge ttier 1963). Tha t is to say , if an individual possesses the c oncept of knowledge , th e n s he ha s th e c a pa c ity to m a ke thi s ju dg me nt. On e wh o f ull y po sse sse s a c o n c e p t n e e d n o t k n o w t h e a p p l i c a t i o n c o n d i t i o n s o f t h e c o n c e p t e x p l i c i t l y. Ideal and Sui table Candid ates for P artic ipation in T hough t Experim ents I n the pre vious section, I talk about both ideal and suitable candida tes for participa tion in thought experiments, but I do not disti ng uish them clear ly . Both a n ideal c a nd ida te fo r p a rt ic ipa tio n in tho ug ht e xpe ri me nts a nd a su ita ble c a nd ida te sh ou ld f ull y possess the conc ept in question, since only those individuals will have reliable intuiti ons a bo ut c a se s in vo lvi ng the c on c e pt. An ind ivi du a l w ho ha s to c on su lt a lin g uis tic a uth or ity a bo ut c e rt a in r e le va nt q ue sti on s a bo ut p os sib le c a se s is no t a su ita ble c a nd ida te fo r a sk ing qu e sti on s a bo ut t a rg e t sc e na ri os if we wa nt r e sp on se s to po ssi ble c a se s to te a c h u s so me thi ng a bo ut o ur c on c e pts or to s e rv e a s e vid e nc e fo r o r a g a ins t a ph ilo so ph ic a l po sit ion . A n id e a l c a nd ida te is o ne wh o h a s a ll o f t he re le va nt r e so ur c e s to arr ive at the r ight a nswer to a canonic al question, including the ability to answer c a no nic a l qu e sti on s c or re c tly wi tho ut e mpi ri c a l in fo rm a tio n. An ind ivi du a l is sti ll a su ita ble c a nd ida te , o n th e oth e r h a nd , e ve n if sh e ne e ds (f or so me tho ug ht e xpe ri me nts ) e mpi ri c a l in fo rm a tio n to a rr ive a t th e ri g ht a ns we r t o th e c a no nic a l qu e sti on . I wi ll illust rate the distinction between a n ideal and a suitable ca ndidate for participa tion in a thought e x periment be low in a discussion about Putnam's Tw in Earth thoug ht experiment (Putnam 1975).

PAGE 134

125 Ag ain, I assume Ry le' s analy sis of ' knowledg e-tha t', but it is not nece ssary to do 19 so to make the point I want to make. I discuss Wi lkes at leng th in 20 Kanuc k 2000. See my discussion of Wil kes in the sec tion of this chapter e ntitled, "Source s of 21 Error and Corre ction." Gendler divides thought e x periments into three differ ent kinds—fac tive, 22 c on c e ptu a l, a nd va lua tio na l—b ut e a c h in vo lve s a kin d o f c ou nte rf a c tua l (Ge ndler 2000, 25). I n a dd iti on to f ull y po sse ssi ng the re le va nt c on c e pts , th e re is a lso a c e rt a in a mou nt o f p ro po sit ion a l kn ow le dg e , o r k no wl e dg e -t ha t, t ha t bo th a n id e a l a nd a su ita ble candida te for pa rticipation in a thoug ht experiment must have: the individual has to be clea r about wha t he or she is be ing a sked to do when pa rticipating in a thoug ht experiment. That is, the individual must understand what a thought e x periment is. She 19 has to know that wha t she is being asked to do is to respond to a que stion about the possible scena rio. She is not being a sked about wha t the individuals in t he thoug ht experiment (if there are any ) would think about the possible scena rio, but, rather , what she thinks about it. Although this see ms to be obvious, W ilkes, who arg ues ag ainst the utilit y of thought e x periments in Re al P e op le : P e rso na l I de nti ty W ith ou t Th ou gh t E x pe rim e nts , 20 for e x ample, see ms to be confuse d about what we are supposed to be doing in a thoug ht e xpe ri me nt (Wilkes 1988). This, I think, is what leads her to arg ue ag ainst their utility . There is evidence that she thinks that when we are conducting a thoug ht experiment wher e ther e ar e individuals behaving in cer tain way s, we a re suppose d to ask about wha t tho se ind ivi du a ls t hin k a bo ut t he po ssi ble sc e na ri o r a the r t ha n w ha t w e do . B ut t ha t is a 21 c on fu sio n. We a re int e re ste d in ou r c on c e pts , n ot s ome oth e r b e ing s' c on c e pts . T hu s, the ind ivi du a l ha s to kn ow tha t sh e is b e ing a sk e d a bo ut h e r o wn c on c e pts a nd tha t sh e is supposed to be imag ining a possible situation. The c andidate ha s to know that the 22

PAGE 135

126 sit ua tio n th a t sh e is p re se nte d w ith is n ot a n a c tua l si tua tio n ( i.e ., it i s ju st a c on c e ptu a lly possible situation). She has to understand, for example, that she ca nnot say that the thought e x periment is not a g ood one bec ause it contra dicts the laws of phy sics. Now that we have de termined to wha t ext ent one must possess a c oncept to be a n ideal ca ndidate for participa tion in a thought experiment, we must determine whe ther a n individual actually possesses the c oncept to the de g ree require d for pa rticipation. Ther e a re tw o w or ri e s th a t ma y c ome up he re : th e ind ivi du a l po sse sse s th e c on c e pt o nly nominally , or the individual possesses a dif fer ent conc ept. The f ormer ma y lead us into a wo rr y a bo ut c ir c ula ri ty wi th r e sp e c t to wh e the r o r n ot w e po sse ss c on c e pts in t his se ns e and how we know that we posse ss conce pts in thi s sense, that is, how we know whethe r the c on dit ion s f or fu ll c on c e pt p os se ssi on a re me t. I t se e ms t ha t th e on ly wa y to deter mine to what extent an individual possesses a conc ept is by having her c onsider possible case s. But, as w e stated a bove, we do not want to ask someone w ho does not antec edently possess the conc ept in question to participate in a thought e x periment beca use her intuiti ons are not reliable. So, the re is a qu e sti on a bo ut h ow we de te rm ine wh e the r s ome on e po sse sse s a conce pt fully . That is, it looks as if to determine whe ther someone possesses the c oncept we a re inter ested in, we may have to a sk her to par ticipate in a thoug ht experiment, but whether her intuitions are r eliable in that thoug ht experiment, we have to know whether sh e po sse sse s th e c on c e pt. We c a n s e e wh e the r s he ha s to or wo uld ha ve to c on su lt a linguistic authority to see whe ther the c oncept a pplies. Assuming that she doe s not or would not, not having to c onsult a linguistic authority is not a sufficient c ondition for kn ow ing tha t a wo rd e xpre sse s a c on c e pt t ha t sh e fu lly po sse sse s. Ho we ve r, the re se e ms to be no re ason to suspect that she doe s not fully possess it. Unlike for nominal

PAGE 136

127 Th e re is a da ng e r t ha t lu rk s h e re . Wh e n th e pa rt ic ipa nt i s a ph ilo so ph e r, sh e is 23 indoctrinated w ith philos ophical theor ies. As a pa rticipant, howe ver, the philosopher must put t hose thing s she lear ned in philosophy aside a nd look for the or dinary individual in herself . po sse ssi on , w e do no t ha ve a su ff ic ie nt c on dit ion fo r w he the r s ome on e po sse sse s a conce pt fully . I n addition, we do not re ally need to test to dete rmine whe ther someone po sse sse s a c on c e pt f ull y . I n p ra c tic e we c a n a ssu me ind ivi du a ls w ho a re fu lly competent use rs of the r eleva nt rang e of voc abular y possess the re levant conc epts, and competenc e is something w e ca n rec og nize usually without recour se to thoug ht experiments. Candid ates for Conducting T hough t Experim ents (Experim enter ) The e x perimente r is the individual who is try ing to f ind conce ptual knowledg e by c on du c tin g a tho ug ht e xpe ri me nt. Th e pr oc e ss o f c on du c tin g tho ug ht e xpe ri me nts is idealized here , for I have se para ted out the role of the experimenter f rom the par ticipant in t his dis c us sio n. Of te n th e e xpe ri me nte r a nd the pa rt ic ipa nt a re the sa me ind ivi du a l, A namely the philosopher who is doing conce ptual analy sis . What ex perimente rs often do is think about how they would answe r questions about a ta rg et sce nario, that is, they suppose that the answ ers they g ive other c ompetent spea kers of the lang uag e would g ive also. The philosopher ty pically asks what she would say , and assumes in doing so that she is like other individuals in the rele vant re spects, so that a c onclusion about a c oncept expressed by a ter m in the public lang uag e is rea ched. That one pe rson may fill two 23 ro le s in c on du c tin g tho ug ht e xpe ri me nts sh ou ld n ot a ff e c t th e re su lts of the pr oc e du re if the re qu ir e me nts fo r c on du c tin g tho ug ht e xpe ri me nts a re me t. H ow e ve r, we sh ou ld s til l important g et clea r on these dif fer ent roles whe n try ing to g et clea r on thoug ht experiments in ge nera l.

PAGE 137

128 Whe n I us e the e xpre ssi on , ' a pr op os e d c on dit ion ' , I me a n a c on dit ion tha t is 24 under c onsideration. What does the experimenter ha ve to be a ble to do in order to c onduct a thoug ht experiment proper ly ? What makes a g ood thought e x perimente r? For my purposes, I wi ll c on tin ue e xpla ini ng the me tho d o f t ho ug ht e xpe ri me nts in t he ide a lize d s e ns e , th a t is , a s if the ro le of e xpe ri me nte r a nd pa rt ic ipa nt w e re se pa ra te d. Th e g oa l of thi s se c tio n is to g e t c le a r o n e xac tly wh a t th e e xpe ri me nte r' s jo b is a nd wh a t kn ow le dg e sh e ha s to have. I n a dd iti on to f ull y po sse ssi ng the c on c e pt i n q ue sti on , th e e xpe ri me nte r m us t have a cer tain fac ility for de signing (that is, setting up a nd car ry ing out) thought experiments. So, the ex perimente r must have more skills t han ar e minimally require d of a par ticipant in a thoug ht experiment. This is because the experimenter ha s to make sure tha t th e ta rg e t sc e na ri o h e is a sk ing the pa rt ic ipa nt t o im a g ine is r e le va ntl y c omp le te ly descr ibed and is a c ase tha t draws on intuitions about the rig ht conce pt and about fe ature s rele vant to the conc ept' s application in which she is intere sted. I n order to give a g ood descr iption of a targ et sce nario, the e x perimente r must be able to tea se apa rt the diffe rent things that mig ht be involved in identify ing w hat fa lls under the conc ept in question. That is, the experimenter must be a ble to descr ibe diffe rent tar g et sce narios involving the same c oncept, some of which conc entra te on ce rtain fe ature s (applica tion conditions) of the conc ept and other s that conce ntrate on othe r fe ature s (applica tion conditions) of the conce pt. We can use thought e x periments to do at lea st four diffe rent thing s, and an e xpe ri me nte r t y pic a lly tr ie s to do a ll o f t he se wh e n in te re ste d in a pa rt ic ula r c on c e pt: (a) to show that a propose d nece ssary condition is necessa ry ; (b) to show that a pr oposed ne c e ssa ry c on dit ion is n ot n e c e ssa ry ; (c ) t o s ho w t ha t a pr op os e d s uf fi c ie nt c on dit ion is 24

PAGE 138

129 I t may turn out that this leads to another wor ry about circ ularity , since 25 deter mining which f eatur es should be we eded out involves c onducting a thoug ht e xpe ri me nt. no t su ff ic ie nt; a nd (d ) t o s ho w t ha t a pr op os e d s uf fi c ie nt c on dit ion is s uf fi c ie nt. I wi ll g ive examples of thoug ht experiments about the conce pt of knowledge neede d to show (a) –(d). This is admittedly a diffic ult project, and some a re e asier to e stablish than oth e rs . E sta bli sh ing su ff ic ie nt c on dit ion s is c le a rl y the mos t di ff ic ult . (a) To show that true be lief is nece ssary for knowle dg e: Suppose that Smit h believes tha t Jone s w ill g e t th e job tha t th e y bo th a pp lie d f or . A s a ma tte r o f f a c t, howeve r, Smith wil l ge t the job. Does Smith know t h a t J o n e s w i l l g e t t h e j o b ? It seems that most participants would say , "no." (b ) To sh ow tha t c e rt a int y is n ot n e c e ssa ry fo r k no wl e dg e : Su pp os e tha t Smi th h a s a true justified belief that he will g et the job that he a nd J ones have applied for , but since Smith has had fa lse beliefs in the pa st, he is not certa in that he will ge t the job. Does Smith know tha t he wi ll g e t th e job ? I t se e ms t ha t mo st p a rt ic ipa nts would say , "y es." (c) To show that true justified be lief is not sufficient for knowledg e: Suppose that Smit h a nd Jone s h a ve a pp lie d f or the sa me job a nd tha t Smi th i s ju sti fi e d in be lie vin g tha t Jone s w ill g e t th e job a nd tha t Jone s h a s 1 0 c oin s in his po c ke t, from which it follows that the man w ho will ge t the job has 10 coins in his pocket. So, Jone s is jus tif ie d in be lie vin g tha t th e ma n w ho wi ll g e t th e job ha s 1 0 c oin s in his pocket. Suppose furthe r that Smith does not know that he, not J ones, will g et the job and he ( that is, Sm ith) has 10 coins in his pocket. Doe s Smi th know who will ge t the job? I t seems that most participants would say , "no." (Ge ttier 1963) (d) To show that true unde fea ted justified belief is suff icient for know ledg e: I will not o f f e r a th o u g h t e xp e r im e n t h e r e , b u t w il l e xp la in w h a t o n e mi g h t l o o k li k e . We may establish sufficie ncy by survey ing a bunch of c ases that show tha t a true undefe ated justified belief is sufficient for knowledg e and a g ree ing tha t the condition is sufficient. Surely there is alway s a possibilit y that there is a defe ater we ha ve not thoug ht of, but it does not look as if there is any way to ge t around tha t. As part of g etting a t or g etting c loser to the nec essar y and suff icient applica tion conditions of a conc ept or tra cing connec tions between c oncepts it is important to weed out the fea tures that ar e mer ely ordinarily corr elated w ith the conce pt(s) in question. 25

PAGE 139

130 For example, consider the f ollowing thoug ht experiment introduced by Wil liams (1970). I t involves the conce pt of pe rso na l id e nti ty : Ca se I: Two per sons, person AA , th e pe rs on a sso c ia te d w ith ps y c ho log ic a lsta te sA , body /brainA and per son BB , th e pe rs on a sso c ia te d w ith ps y c ho log ic a lsta te s01 B , body /brainB , a re tol d a t t a bo ut a n o pe ra tio n th e y wi ll u nd e rg o a t t the re su lt 2 of wh ic h is tha t a t t brainB is a sso c ia te d w ith the ps y c ho log ic a l st a te s p re vio us ly associate d with AA , and bra inA is associated w ith the psy cholog ical states previously associate d with BB . Two per sons emerg e— AB and BA . I mag ine 0, fu rt he r t ha t a t t befor e AA underg oes the proc edure , she is told that either AB or BA will be g iven a lar g e sum of money and the other will be tortured. AA is then asked to make a selfinterested de cision as to whether the re war d should be g iven 2 to AB or BA a t t . What would it rational for AA to choose? Ca se II: Two per sons, person AA , th e pe rs on a sso c ia te d w ith ps y c ho log ic a lsta te sA , body /brainA and per son BB , th e pe rs on a sso c ia te d w ith ps y c ho log ic a l01 sta te sB , body /brainB , a re tol d a t t a bo ut a n o pe ra tio n th e y wi ll u nd e rg o a t t the 2 re su lt o f w hic h is tha t a t t brainB is associated w ith the psy cholog ical states previously associate d with AA , and bra inA is associated w ith the psy cholog ical states pre viously associate d with BB . Two per sons emerg e— AB and BA . 0 I ma g ine fu rt he r t ha t AA is t old by the su rg e on a t t tha t she will be tortured a fter 2 the pr oc e du re , a t t , but that during the torture she will have no memory of being told so. Further more, she will not have any memories of her past; rather , she wi ll have a new se t of impressions of her past. Wit h respe ct to Case I , Wil liams say s that it would be rational for AA to choose AB to ge t the re wa rd . T his is s up po se d to be a n in dic a tio n th a t w e thi nk pe rs on a l id e nti ty lie s in one' s psy cholog y . Wit h respe ct to Case I I , Wil liams arg ues that the pe rson associa ted 0 wi th p sy c ho log ic a lsta te sA , body /brainA a t t wi ll b e dis tur be d a nd c on c e rn e d th a t bo th 2 her bo dy a nd min d w ill be ha rm e d. Th a t BA a t t has no memory of having been told the 1 ou tc ome of the pr oc e du re be fo re ha nd a nd kn ow s a t t t h a t b o d yA wi ll b e a sso c ia te d w ith 2 dif fe re nt b ra insta te s a t t s e e m s n o t t o a f f e c t t h e w a y she re a c ts.

PAGE 140

131 Sur e ly the e xpe ri me nte r c ome s w ith so me ide a s a bo ut w ha t is re le va nt w ith 26 re sp e c t to a c on c e pt' s a pp lic a tio n c on dit ion s. Ho we ve r, the e xpe ri me nte r i s f a lli ble . T his is not a problem bec ause the experimenter c an test to see w hether the fe ature s she thought we re r eleva nt to the application of the c oncept a re. I n fac t, having ideas a bout what is rele vant with respe ct to a conc ept' s application conditions is useful in desig ning thought e x periments. I am talking about conc epts as if they have ne cessa ry and suff icient applica tion 27 c on dit ion s. Th is t ho ug ht e xpe ri me nt w ou ld g o th e sa me wa y (i n th e re le va nt r e sp e c ts) for someone like Wit tge nstein (1958) who thinks that all (or a t least most) conce pts are fa mil y re se mbl a nc e c on c e pts . Ge nd le r ( 19 98 ) a rg ue s th a t in tho ug ht e xpe ri me nts inv olv ing pe rs on a l id e nti ty 28 le a vin g ou t th e a c c ide nta lly c or re la te d f e a tur e s is wh a t ma ke s th e m mi sle a din g . I dis c us s this in detail in Kanuc k (2002). Clearly things like same ness of phy sical appe ara nce, voic e, etc . in all actual c ases tend to be cor rela ted with samene ss of person ove r time. We can c all these fe ature s 26 epistemic indicators since we of ten make judg ments about persons on the ba sis of them. Howeve r, no one w ould arg ue that same ness of phy sical appe ara nce or voice is nec essar y or suffic ient for pe rsonal identity . I t is im portant in ce rtain thoug ht experiments, then, 27 to leave me ntion of these ac cidental f eatur es out of the de scription of the c ase so that the participa nt does not g et sway ed by those ac cidental f eatur es. Notice that in the a bove 28 tho ug ht e xpe ri me nt, the re is n o me nti on of a c c ide nta l f e a tur e s th a t a re or din a ri ly corr elated w ith the conce pt in question. I t is im portant for the experimenter to lea ve out the fe ature s that are mere ly acc identally corr elated w ith the nece ssary and suff icient a pp lic a tio n c on dit ion s o f t he c on c e pt f ro m th e de sc ri pti on of the ta rg e t sc e na ri o. Th is i s so the participa nt is sure to make judg ments solely based on w hat is in the running for be ing a su ff ic ie nt c on dit ion fo r a pp lic a tio n o f t he c on c e pt i n q ue sti on . I t se e ms t ha t in the c a se of pe rs on a l id e nti ty , th e on ly c on te nd e rs fo r a pp lic a tio n c on dit ion s a re sa me ne ss of body and same ness of psy cholog y . The e x perimente r must at least know this to be able to come up with a thoug ht experiment that is relevant to g etting c lear on the conc ept

PAGE 141

132 For example, we ca n g ive a thoug ht experiment about persona l identity wher e 29 we include in the descr iption of the case one individual who has the sa me hair c olor and voice but diffe rent body and diffe rent psy cholog ical states, e tc. and a nother who ha s the same psy cholog ical states, e tc., and a sk who is identical with the ea rlier individual. in question. Getting r id of all of the a ccide ntally corr elated f eatur es on which a participa nt may base he r judg ment is difficult, and we may have to r un sever al thoug ht experiments to make sure tha t she does not base her judg ment on one or more of them. An e xpe ri me nte r c a n g e t r id o f a c c ide nta lly c or re la te d f e a tur e s o f a c on c e pt ( or if she is inclined to g et clea r on wha t fea tures ar e ac cidentally corr elated w ith a conce pt ) 29 by doing se vera l thought experiments. Fir st, she can de scribe one possible case that leave s the presume d acc identally corr elated f eatur e out and lea ves the pre sumed ne c e ssa ry or su ff ic ie nt c on dit ion in a nd a no the r t ha t in c lud e s th e pr e su me d a c c ide nta lly corr elated f eatur e, but leave s the presume d nece ssary or suffic ient condition out of the descr iption. Every thing other than that should remain the sa me in the desc ription. Then, the experimenter should c ompare the re sults of eac h targ et sce nario. I f the re sult of the first targ et sce nario is that the conc ept does a pply and the r esult of the sec ond is that the c on c e pt d oe s n ot a pp ly , th e n w e c a n c on c lud e tha t th e fe a tur e is, in f a c t, a n a c c ide nta lly corr elated one . Alterna tively , if the re sult of the first targ et sce nario is that the conc ept does not apply and the r esult of the sec ond targ et sce nario is that it does apply , then we may conclude that the pre sumed ac cidental f eatur e is a suff icient condition for a pp lic a tio n o f t he c on c e pt. I t ta ke s a lon g tim e to g o th ro ug h a ll o f t he po ssi bil iti e s, bu t it is p la us ibl e tha t w e c a n e ve ntu a lly str ip a wa y a ll o f t he re le va nt a c c ide nta l c or re la tio ns in this way . I n prac tice ther e will turn out to be quite a lot of initial ag ree ment on the irrele vance of many fea tures, and thus tests nee d to be conduc ted only when que stions arise a bout them.

PAGE 142

133 A g ood experimenter must also have more skills that are not r equire d of the participa nt. Whereas the pa rticipant only has to know wha t a thoug ht experiment is and wh at s h e i s b ei n g as k ed t o d o , t h e e x p er i m en t er h as t o d o a n u m b er o f a d d i t i o n al t h i n gs a nd ha ve a t le a st t hr e e a bil iti e s, a ll o f w hic h r e qu ir e a fu ll g ra sp of the c on c e pt i n question plus a fac ility for de signing thought e x periments: (a ) She has to be imag inative e no ug h to c ome up wi th i ma g ina ry c a se s a bo ut t he c on c e pt; (b ) S he ha s to be a ble to de te rm ine wh ic h p os sib le c a se s a re mos t r e le va nt i n g e tti ng c le a r o n th e c on c e pt i n question; and (c) She has to be a ble to g ive a r eleva ntly complete a nd car eful de scription of a ta rg et sce nario. I n th e ne xt se c tio n, I wi ll g ive e xamp le s o f b a d th ou g ht e xpe ri me nts on my fra mework f or thinking a bout thought experiments and introduc e some skeptica l views about thoug ht experiments. Then, I will ex plain how we c an make , and cor rec t, mist akes in perfor ming thoug ht experiments. Sources of Error and Corre ction I have introduc ed a f rame work for thinking about thoug ht experiments and g ave some examples of what I consider to be g ood thought e x periments. I also g ave conditions for par ticipation in thought experiments, which included a discussion about what it takes to possess a c oncept. I n this section, I will give some examples of bad thought e x periments and e x plain the way s in which we c an make mistakes in perfor ming wh a t w ou ld b e c on sid e re d g oo d th ou g ht e xpe ri me nts . T he n, I of fe r w a y s to c or re c t th os e mistakes. F ew Bad P hilos ophi cal Thou ght Experim ents Now that we have se en some e x amples of g ood thought e x periments, let us c on sid e r s ome ba d o ne s. A b a d th ou g ht e xpe ri me nt i s o ne wh ic h w e sh ou ld n ot u se to

PAGE 143

134 P er h ap s i t wo u l d b e u s ef u l i n s u ch a c as e t o p ro v i d e b ef o re an d af t er d ra wi n gs 30 a nd a llo w t he pa rt ic ipa nt t o c omp a re the m. tell us things about our c oncepts. I do not in this section attempt to g ive conditions for a thought e x periment to be a bad one; I just want to show the kinds of things that may make a thought e x periment a bad one. Some philosophers w ho are skeptical a bout the utilit y of thoug ht experiments, I suspect, think that all thought e x periments ar e bad one s. I have e x plained why I do not think thi s is true ea rlier, whe re I g ive examples of g ood thought e x periments and g ive some conditions for wha t it takes for a thought e x periment to be a g ood one. Be fore g iving a text ual example of a ba d thought e x periment, I will give a n e xamp le of wh a t w ou ld b e a ve ry sim ple , b ut b a d, tho ug ht e xpe ri me nt: Suppose that there is a n individual who has a ve ry few hairs on top of his hea d, bu t ma ny ha ir s a ro un d th e ba c k o f h is h e a d a nd on the sid e s o f h is h e a d. Th is individual sees an a d in the paper for a solution t hat is supposed to cure baldness. He or ders the solution and uses it ac cording to the direc tions, but he only g rows some new ha irs. The individual sues the c ompany for f alse a dvertising . 30 Ca noni c al q ue st ion : Sho uld he wi n th e c a se or no t? Th is t ho ug ht e xpe ri me nt i s b a d f or a t le a st t wo re a so ns . F ir st, the de sc ri pti on of thi s targ et sce nario is too vag ue. I t is unclear how many hairs ar e ver y few and how many are the " some" hairs that he g rows af ter using the produc t. Second, there is an auxil iary presumption here (one tha t may not be war rante d) that the c oncept of ba ldn e ss is a deter minate conc ept, or, ra ther, that it is a conc ept. This is a ca se whe re the intuiti ons we g et should allow us to make ce rtain infer ence s: if the participa nt answer s y es, then we should be able to infe r that she still thinks t he individual is bald; if the participa nt a ns we rs no , th e n w e sh ou ld b e a ble to i nf e r t ha t sh e no lon g e r t hin ks the ind ivi du a l is bald. I t seems that in such a c ase, how ever , it is i ndeter minate whethe r the individual

PAGE 144

135 sh ou ld w in h is c a se , s inc e it i s in de te rm ina te wh e the r t he fe w n e w h a ir s ma ke him unbald. So, we should not forc e the pa rticipant to make a definitive judg ment about the case . That is, there are case s where we should not try to force a def initive judgment on a participa nt when we oug ht to allow for indeter minacy . Or, if it is determinate whether he sh ou ld w in t he c a se , h is w inn ing do e s n ot d e pe nd on wh e the r h e is b a ld, bu t on so me o t h e r f a c t o f t h e c a s e . E v e n i f w e s u p p o s e — a s w e n e e d n o t — t h a t c o u r t c a s e s a r e a l w a ys decide d on matters of f act, the ma tter of f act mig ht be some other f act—not the f act that the man is bald or not. Wil kes (1988) arg ues that instead of considering possible case s when we do philosophy (in particula r, per sonal identity ), we should consider a ctual ca ses. One of the way s that she ar g ues ag ainst thought e x periments is by pointing out a c ommon failing of them. She arg ues that we cannot g et a g ood experimental set-up (i.e ., a g ood descr iption of the c ase) . There is one thoug ht experiment that W ilkes uses as an e x ample in which we a ll spli t like amoebae : Consider next one of the fa miliar thought e x periments to do with persona l ide nti ty : th a t w e a ll s pli t li ke a moe ba e . I t is ob vio us ly a nd e sse nti a lly re le va nt to the purposes of this thought experiment to k no w s uc h th ing s a s: h ow of te n? I s it predic table? Or sometimes pre dictable a nd sometimes not, like dy ing? C an it be induced, or preve nted? J ust as obviously , the bac kg round society , ag ainst which we se t the phenomenon, is now my sterious. Does it have such instituti ons as marria g e? How would that wor k? Or univer sities? I t would be difficult, to say the least, if univer sities doubled in siz e eve ry few day s, or wee ks, or y ear s. Are preg nant women de barr ed fr om split ting? The e nti re ba c kg ro un d h e re is inc omp re he ns ibl e . Wh e n w e a sk wh a t we would say if this happened, w ho, now, a r e ' w e ' ? (Wilkes 1988, 11) Be fore we g et to other proble ms with W ilkes' s diag nosis of this thought experiment, we should notice that this description of a ta rg et sce nario, if she wer e to use it in perf orming a thoug ht experiment, is si mply incomplete. I t is unclear w hat the par ticipant is supposed to be having intuiti ons about (i.e., it is not at all clea r wha t we should be imag ining) .

PAGE 145

136 There is no question and, rea ding a s char itably as we can, no c lues about wha t kind of question the participa nt should answer. The questions that she does ask se em to have nothing a t all to do with our conce pt of pe rso na l id e nti ty . We cannot eve n rew ork wha t wo uld be the ta rg e t sc e na ri o s o th a t w e c a n g e t a qu e sti on ou t of it, y e t sh e g oe s o n to c ri tic ize it. Th e fo llo wi ng is w ha t sh e sa y s a bo ut t he tho ug ht e xpe ri me nt: Th e po int he re is, I thi nk , c e ntr a l. T ho ug ht e xpe ri me nts a re , li ke a ll e xpe ri me nts , underta ken for a spec ific purpose . Thoug ht experiments to do with personal identity have, a s their offic ial ambition, the job of reve aling the hea rt of our current , present , notions of what it is to be a person. B ut the only one discussed s o fa r— t h e a m o eb as t y l e s p l i t — s ee m s t o l ac k t h e c ap ac i t y t o d o an y s u ch t h i n g, a nd fo r t his re a so n: in a wo rl d w he re we sp lit lik e a moe ba e , e ve ry thi ng e lse is g oin g to b e so un ima g ina bly dif fe re nt t ha t w e do no t kn ow wh a t c on c e pts wo uld re ma in ' fi xed ,' pa rt of the ba c kg ro un d; w e ha ve no t f ill e d o ut t he re le va nt d e ta ils of thi s ' po ssi ble wo rl d, ' e xce pt t ha t w e kn ow it c a nn ot b e muc h li ke ou rs . B ut i f we c annot know that, then we cannot a ssess, or der ive conc lusions from, the thought e x periment. (Wilkes 1988, 12) Wilk e s sa y s h e rs e lf tha t th ou g ht e xpe ri me nts a re " un de rt a ke n f or a sp e c if ic purpose," but it is entirely unclea r for what purpose the amoeba one would be un de rt a ke n. F ur the rm or e , it is u nc le a r t ha t su c h a wo rl d, if re le va ntl y c omp le te ly descr ibed, would be un ima g ina bly dif fe re nt f ro m ou rs . T he ma in p ro ble m w ith thi s thought e x periment, if pe rfor med, is that it is not even r eally a thoug ht experiment at all. F ur the rm or e , Wil ke s ma ke s th e mis ta ke of su g g e sti ng tha t if we we re to p e rf or m th is tho ug ht e xpe ri me nt t ha t we would be the individuals in the targ et sce nario. I suspect that this is why she thinks we would have to answer so many questions about the bac kg round conditions in the description of the c ase. So, if per formed, this thoug ht experiment would be a ba d one, and, thus would not ef fec tively point out failings of thoug ht experiments for at lea st four re asons: (a) the targ et sce nario would be underde scribed; (b) it would be e ntirely unclea r wha t the participa nt is being a sked about; (c ) the que stion would not fit into t he ca nonical questions I prese nted ea rlier in the c hapter ; and (d) w e

PAGE 146

137 wo uld no t be a sk e d a bo ut o ur c on c e pts , b ut a bo ut s ome oth e r b e ing s' c on c e pts (t ha t is , t h e c o u n t e r f a c t u a l q u e s t i o n t h a t W i l k e s ' s a s k s d o e s n o t h o l d f i x e d o u r c o n c e p t s ) . In addition, the way the targ et sce nario would be descr ibed, it is clear tha t we would be interested in something , but we would not be a ble to take wha t the experimenter a sks seriously . All of this can be c onfusing and misleading , even to idea l participants, so the example she uses does not do wha t she purports it does. Gale ( 1991) see ms to point out the same fa ilings of thoug ht experiments that do not work either : What is perverse about these sc ience -fiction thoug ht-experiments is that they tr a ns po rt us , a lon g wi th o ur pr e se nt l a ng ua g e -g a me s a nd the ir fo rm s o f l if e , in to the counter -fa ctual wor ld. And it is claimed that in such a wor ld we fa ce c ounterexamples to our analy sis of persona l identity or at lea st undecidable case s. What they fail to rea liz e is that in this world we would not want to play our old persona l identity lang uag e-g ame, since there would be no point or value in doing so: the e mpi ri c a l pr e su pp os iti on s a re no t r e a lize d . . . [T]he re no lon g e r i s n or ma tiv e ly rule-g overne d prac tice of ide ntify ing a nd reide ntify ing pe rsons. (Ga le 1991, 301) Ag ain, this just s eems to be c onfused. I n thought e x periments, we are not supposed to be transported to the counter -fa ctual wor ld. Rather, we are supposed to be looking at the counter -fa ctual wor ld, so-to-speak, f rom the outside and se e whe ther the c oncept a pplies or fa ils to apply , or whe ther it is indeterminate tha t it does, to the possibl e sce nario. Fur thermore , Gale e x plicitly talks about the lang uag e we would use in that situation, which would not be our s, and states that we would not want to use ou r o wn c on c e pts . I do not in this section pre tend that I have c overe d all of the wa y s in which a thought e x periment c an be a bad one. I mere ly want to show that there are bad thoug ht experiments in some of the literature . We should make sure tha t when using thought e xpe ri me nts to a rr ive a t ph ilo so ph ic a l po sit ion s, we a re us ing g oo d o ne s.

PAGE 147

138 Skeptical View s About the Utility of T hough t Experim ents Since a number of philosophers ar e skeptica l about the utility of thoug ht experiments, y et I have a rg ued that thoug ht experiments play a ce ntral role in doing philosophy , it is i mportant in this section to ge t clear on why those philosophers have been ske ptical and to explain why their skepticism is unwarr anted. I f some philosophers use thoug ht experiments incorrec tly and that is what make s them skeptical, we should not there fore stop conducting thought e x periments. Ea rlier, I g ave some nece ssary c on dit ion s f or c on du c tin g tho ug ht e xpe ri me nts c or re c tly , v iol a tio n o f w hic h le d to sk e pti c ism . G ivi ng the m sh ou ld h e lp a lle via te so me of the sk e pti c ism . F ir st, I wi ll dis c us s a wo rr y a bo ut c ir c ula ri ty tha t on e ma y ha ve a nd a re la te d w or ry . Se c on d, I wi ll discuss a worr y about the complete ness of the de scription of a ta rg et sce nario. Third, I w i l l d i s c u s s t h e w o r r y t h a t w e m a y n o t k n o w w h a t t o s a y i n a b n o r m a l s i t u a t i o n s . In Chapter 4 of this diss erta tion, I will deal with issues about skepticism about conc eptual a na ly sis in g e ne ra l. I n the sec tion of this chapter e ntitled, "Suitable Candidates a nd Concept Pos se ssi on ," I ra ise d a wo rr y a bo ut t ho ug ht e xpe ri me nts tha t is re la te d to the de g re e in which a pa rticipant in a thoug ht experiment should possess the concept in question: the individual may possess the conc ept nominally . I f an individual possesses the c oncept no min a lly , th e n s he is n ot a su ita ble pa rt ic ipa nt f or the tho ug ht e xpe ri me nt, bu t it se e ms that we c annot be sure whether an individual possesses a c oncept f ully unless we a sk the individual to consider thoug ht experiments, and one who doe s not fully possess the conce pt in question will not have r eliable intuitions when consider ing ta rg et sce narios. F ur the rm or e , a nd a lon g the sa me lin e s, it m a y be un c le a r w he the r w e po sse ss c on c e pts fully in the first place , so it looks as if we have to do thought experiments to deter mine

PAGE 148

139 whether we do. I t seems that this might lead to a worry about circ ularity : we do thoug ht experiments to learn thing s about our conc epts, but to learn a bout our conc epts by doing thought e x periments, we have to do a dditional thought experiments. As I explained a bo ve , th is i s n ot a pr ob le m be c a us e we c a n d e te rm ine wh e the r s ome on e on ly no min a lly possesses a c oncept if she has to defe r to a ling uistic authority . I n addition, in practice we c an assume tha t an individual who is a fully competent use r of the r eleva nt vocabula ry fully possesses the c oncept, a nd we c an re cog nize competence without using thought e x periments. Of course , the assumption may be fa lse, but then this will s how up either in the subjec t not knowing wha t to say or in devianc es betwe en her intuiti ons and those of other speake rs of the la ng uag e. Th e re is a re la te d w or ry tha t on e mig ht h a ve a bo ut t ho ug ht e xpe ri me nts tha t is muc h mo re se ri ou s: i f a n in div idu a l po sse sse s a dif fe re nt c on c e pt t ha n th e c on c e pt i n qu e sti on , b ut o ne tha t mi mic s it in c e rt a in w a y s ( fo r e xamp le , it ma y be ha ve lik e it i n cer tain way s), it may be diffic ult, if not im possible, to determine whe ther the individual fully possesses the re levant conc ept. That is, it may be impossible to determine wha t conce pt the individual possesses. I f we cannot de termine whe ther the pa rticipant possesses the re levant conc ept, then we cannot r ely on her intuitions about targ et scena rios that involve the re levant conc ept, for w e ar e not sure w hat conc ept her intuiti ons are about. Consider for example, Quine' s (1960) thesis for the indeter minacy of transla tion, which he illustrates throug h his ga vag ai example: [M] anuals for translating one lang uag e into another can be set up in diverg ent wa y s, a ll c omp a tib le wi th t he tot a lit y of sp e e c h d isp os iti on s, y e t in c omp a tib le wi th o ne a no the r. I n c ou ntl e ss p la c e s th e y wi ll d ive rg e in g ivi ng , a s th e ir respe ctive tra nslations of a sentenc e of the one lang uag e, sente nces of the other lang uag e which sta nd to eac h other in no plausible sort of e quivalence howeve r loose. (27)

PAGE 149

140 I t seems that this thesis can ea sily be tra nslated into a worr y about deter mining wha t conce pt it is that a potential par ticipant possesses: there could be a number of dif fer ent possible conce pts that one may possess g iven her competenc y with the rele vant vocabula ry , that is, give n the totality of her speec h dispositi ons. Quine' s g avag ai example is supposed to show that we ca nnot translate the la ng uag e of " untouched people," which has no r esemblanc e to any lang uag e we know of, be cause there will be a number of e qually acc eptable tr anslations g iven the individual' s behavior. Similarly , one may arg ue that we cannot de termine whe ther a n individual possesses one conc ept, C , rathe r than a nother, c losely rela ted conc ept, C ' . Th e qu e sti on no w i s w he the r t his kin d o f o bje c tio n s ho uld le a d u s to sk e pti c ism about the utility of thoug ht experiments. Aga in, in practice , it seems that g iven the public lang uag e, it is safe to a ssume that if an individual is competent in the re levant vocabula ry , we c an assume tha t she possesses the r eleva nt conce pts. One wa y to ensure that the par ticipant possesses the re levant conc ept is to ask more than one individual for her intuitions about targ et sce narios and to a sk eac h individual to im ag ine seve ral tar g et scena rios involving the r eleva nt conce pt. One mig ht stil l arg ue, howe ver, tha t even if w e do that, it is st ill possi ble for the individual to possess a differ ent conc ept than the one we are interested in. The problem with this kind of objection is that it s eems that we might raise this ag ainst the possibili ty of having meaning ful conve rsations as we ll, but we are no t sk e pti c a l a bo ut t he me a nin g fu lne ss o f o ur c on ve rs a tio ns wi th o the r i nd ivi du a ls. Another w orry that one mig ht have a bout the utilit y of thoug ht experiments is t hat the re is n o p ri nc ipl e d w a y of c ho os ing pa rt ic ipa nts . We c a nn ot c ho os e a pa rt ic ipa nt u nti l we f irst know that the individual possesses the conc ept in question. To find that out, we have to f ind out whether or not the individual gives the rig ht answer s. To find out

PAGE 150

141 whether the individual ge ts the rig ht answer s, we ne ed to have the rig ht answer s alrea dy . B ut w e c a nn ot g e t th e ri g ht a ns we rs un til we do tho ug ht e xpe ri me nts . So , it loo ks a s if we ha ve a r eal pr oblem here : we ca nnot know the applica tion conditions unt il we do the tho ug ht e xpe ri me nts ; w e c a nn ot d o th e e xpe ri me nts un til we c ho os e a su ita ble participa nt; and we c annot choose a suitable pa rticipant until we know the a pplication c on dit ion s. Th is w or ry a bo ut c ir c ula ri ty ma y le a d to a pr ob le m in de te rm ini ng wh o is a suitable participa nt without the philos opher r efle cting on his own theory about the conce pt in question. S ince a n individual has no worry about his own conc epts, he c an alway s use himself and not wor ry about whethe r it is the rig ht conce pt. Alternatively , the e xpe ri me nte r c a n a pp e a l to his ow n p os se ssi on of c on c e pts a nd us e so me thi ng a kin to inductive evidenc e that the pa rticipant has the sa me conc ept. I say above tha t when an e x perimente r asks someone to consider a targ et sce nario, the desc ription of the targ et sce nario must be ca ref ul and re levantly complete. One might claim that it is difficult to know when a de scription is releva ntly complete or how complete a rele vantly complete de scription has to be. The only descr iption of a clea r and c omp le te de sc ri pti on is a c omp le te de sc ri pti on of wh a t is re le va nt t o th e qu e sti on tha t is asked, a nd a desc ription that does not include irre levant or e x tra de tails that might confuse or influenc e a pa rticipant' s intuit ions. Ag ain, deter mining whe ther a descr iption is a ca ref ul and complete one may take doing additional thought e x periments and evalua ting pa rticipants' intuiti ons. Furthe rmore, the experimenter is supposed to be skilled at coming up with descriptions of imag inary case s that will actually draw on a pa rt ic ipa nt' s in tui tio ns .

PAGE 151

142 This kind of objection also appe ars in 31 (Wilkes 1988). Th e fi na l ob je c tio n th a t I wa nt t o a dd re ss i n th is s e c tio n is on e tha t a pp e a rs in Wittg e ns te in' s Philosophical Investigations : 31 I t is only in normal ca ses that the use of a wor d is clear ly presc ribed; we know, are in no doubt, what to say in this or that case. The more a bnormal the c ase, the more doubtful it become s what we are to say . And if thing s were quite differ ent fr om w ha t th e y a c tua lly a re . . . th is w ou ld m a ke ou r n or ma l la ng ua g e -g a me s lo se their point. (Witt g enstein 1958, §142) The objec tion that W ittgenste in raises he re is an inter esting one: there is no clear rea son we should pre sume that a wor d has dete rminate a pplication conditions in unusual cases. I f we idealize lang uag e by assuming tha t words do have deter minate applica tion conditions, then we will be led a stray . Since thoug ht experiments primarily involve me re ly po ssi ble c a se s, y e t w e sh ou ld n ot p re su me tha t ou r t e rm s h a ve de te rm ina te a pp lic a tio n c on dit ion s in un us ua l c a se s, it l oo ks a s if tho ug ht e xpe ri me nts pla us ibl y wi ll of te n le a d u s a str a y . O ur ru le s f or us ing te rm s d o n ot a pp ly to m e re ly po ssi ble c a se s, so a ny int uit ion s w e ha ve a bo ut t he m w ill be mis le a din g . H ow e ve r, it d oe s n ot l oo k a s if such an obje ction is so devastating . All we have to do is gua rd ag ainst taking the ide a liza tio n o f l a ng ua g e too se ri ou sly by ke e pin g in m ind (p a rt ic ula rl y in u nu su a l c a se s) that we a re ide alizing things a bit in order to make some prog ress. I n this section, I have only addre ssed objections to the utility of thoug ht experiments, all of which c an be f airly easily overc ome. Since thoug ht experiments play a c e ntr a l r ole in c on c e ptu a l a na ly sis , th e se ob je c tio ns ma y a lso be ra ise d a g a ins t c on c e ptu a l a na ly sis . I n Ch a pte r 4 , I wi ll a dd re ss o bje c tio ns a g a ins t c on c e ptu a l a na ly sis in ge nera l.

PAGE 152

143 An example of a thoug ht experiment where I think this kind of mistake is being 32 made is in (Wi lke s 1 98 8, 11 ff ), wh e re sh e inc omp le te ly de sc ri be s a ta rg e t sc e na ri o s o t h at we d o n o t ev en k n o w w h at we ar e s u p p o s ed t o b e i m agi n i n g. P roper P erform ance of Th ought Exp erim ents and Possible Sou rce s of E rror Thoug ht experiments that are use d to rea ch metaphy sical conc lusions mus t be done prope rly ; a thoug ht experiment that is not done properly will likely g et us confuse d or mis le a din g re su lts or bo th, a nd , th us , a n im pr op e rl y pe rf or me d th ou g ht e xpe ri me nt i s a source of er ror. I sug g est the following as conditions under w hich a thoug ht experiment is properly perf ormed: A t ho ug ht e xpe ri me nt i s properly performe d if f ( a ) t he ta rg e t sc e na ri o is rele vantly completely and ca ref ully descr ibed, (b) the participa nt is a competent speake r of the la ng uag e that the c oncept in question is expressed in, (c) the pa rt ic ipa nt ful ly possesses the c oncept in question, (d) the par ticipant understands he r t a sk a s a pa rt ic ipa nt i n a tho ug ht e xpe ri me nt, a nd (e ) t he e xpe ri me nte r m e e ts all of the c riteria a suitable participa nt must m eet a nd has a f acility for de signing tho ug ht e xpe ri me nts . Th e pe rf or ma nc e of a ny tho ug ht e xpe ri me nt t ha t do e s n ot m e e t th e se fi ve c on dit ion s is no t a pr op e r o ne , a nd , th us , it s r e su lts sh ou ld n ot b e us e d a s e vid e nc e fo r o r a g a ins t a philosophical position. The philosopher c onducting a thoug ht experiment is responsible for de signing , or se tti ng up , th e ta rg e t sc e na ri o. I f t he e xpe ri me nte r d oe s n ot g ive a re le va ntl y c omp le te descr iption of the targ et sce nario, then the participa nt may understanda bly make a mistake in answer ing the canonic al question. Alterna tively , if the experimenter 32 includes extra details that ca n be conf using to the pa rticipant, she may answe r the

PAGE 153

144 An example of a thoug ht experiment where I think there a re thing s in the 33 descr iption of the targ et sce nario that may confuse a par ticipant is in (Will iams 1970, 181ff) . I arg ue in (Ka nuck 2002) tha t Gendler mishandles Wil liams's thoug ht e xpe ri me nts be c a us e sh e fa ils to r e c og nize on e of the ta rg e t sc e na ri os a c tua lly presuppose s an answe r to the ca nonical question in arg uing tha t we g ive inconsistent a ns we rs to t he c a no nic a l qu e sti on s in tw o d if fe re nt d e sc ri pti on s o f a pp a re ntl y the sa me targ et sce nario. I t is, of course, some times difficult to determine whic h fea tures ar e ac cidental 34 corr elates a nd which ar e nec essar y corr elates of a conc ept. B ut it can be done by doing a dd iti on a l th ou g ht e xpe ri me nts . Se e , f or e xamp le , B ur g e ' s a rt hr iti s c a se wh e re the pe rs on in t he ta rg e t sc e na ri o 35 mis ta ke nly thi nk s th a t he ha s a rt hr iti s in his thi g h. Th a t pe rs on is n ot a su ita ble pa rt ic ipa nt i n a tho ug ht e xpe ri me nt (B urg e 1979). Gale se ems to be conf used about wha t we ar e supposed to do in thoug ht 36 e xpe ri me nts . I n p a rt ic ula r, he se e ms t o th ink tha t w e a re su pp os e d to pu t ou rs e lve s in to the targ et sce nario, but not consider whether our conce pts apply to the situation (G a le 1991, 301). canonic al question incorr ectly . For example, if an experimenter mentions in the targ et 33 scena rio fea tures that ar e ac cidentally corr elated w ith the application conditions of the c on c e pt i n q ue sti on , th e pa rt ic ipa nt m a y a ns we r t he c a no nic a l qu e sti on ba se d o n th os e a c c ide nta l f e a tur e s o f t he c on c e pt r a the r t ha n it s a pp lic a tio n c on dit ion s. 34 Assuming that the e x perimente r g ives a re levantly complete de scription of the targ et sce nario a nd leave s out acc identally corr elated f eatur es, there are still err ors that the pa rt ic ipa nt c a n ma ke . F ir st, a pa rt ic ipa nt w ho do e s n ot p os se ss t he c on c e pt i n qu e sti on wi ll l ike ly a ns we r a c a no nic a l qu e sti on inc or re c tly be c a us e sh e is u na ble to a ns we r m a ny of the qu e sti on s o ne mig ht a sk a bo ut w he the r t he c on c e pt a pp lie s, fa ils to apply , or whe ther it is indeterminate tha t it does. Second, a pa rticipant may answe r a 35 canonic al question incorr ectly if she does not know w hat she is supposed to be doing or is uncoopera tive when a sked to participa te in a thoug ht experiment. Third, a pa rticipant 36 ma y ma ke a mis ta ke in a ns we ri ng a c a no nic a l qu e sti on if sh e po sse sse s th e c on c e pt i n question, but she is not pay ing c lose enoug h attention to the targ et sce nario, or if she does

PAGE 154

145 not have e noug h additional resour ces to make judgme nts about the conc ept' s application (e.g ., if she does not possess a dditional concepts that a re r equire d to answer the ca nonical qu e sti on c or re c tly ). I f a ny of the se e rr or s a re ma de in a tho ug ht e xpe ri me nt, the re su lts do not y ield results that we should use to dra w metaphy sical conc lusions; t he thoug ht experiment has not been pr operly perf ormed. I t is possi ble for a ll of the conditions for prope r per formanc e to be met, y et for a participa nt's intuitions not to s erve as g ood evidenc e for or ag ainst a philosophical po sit ion . O nly wh e n th e c on dit ion s f or pr op e r p e rf or ma nc e a re me t and t h e p ar t i ci p an t 's answe r counts as g ood evidenc e ca n we sa y that a thoug ht experiment has been su c c e ssf ull y pe rf or me d. A p a rt ic ipa nt' s in tui tio n a bo ut a ta rg e t sc e na ri o in a pr op e rl y perf ormed thoug ht experiment may not serve a s g ood evidenc e for a philosophical position when a par ticipant fails to be sensitive in a c erta in way to the rele vance of ba c kg ro un d in fo rm a tio n in a ns we ri ng a c a no nic a l qu e sti on . I wi ll e xpla in t his in d e ta il i n my analy sis of the Twin Ear th thought experiment and the zombi e thoug ht experiment. Now that we are clea rer about the kinds of mistakes we can ma ke in per forming t h o u gh t ex p er i m en t s an d h o w w e c an co rr ec t t h em , we ca n m o v e o n t o ev al u at i n g a particula r thoug ht experiment to see how my fra mework f or thinking a bout thought e xpe ri me nts c a n b e pu t to wo rk . I n th e ne xt se c tio n, I wi ll i ntr od uc e Put na m' s T wi n Earth thoug ht experiment and a va riation of it to illus trate a case in which a thoug ht experiment is properly , but not successfully , perf ormed (Putnam 1975). I n doing this, I will revisit the char acte rization of full conce pt possession I prese nted ea rlier, a nd then e xpla in h ow the kn ow le dg e g ra nte d b y su c h c on c e pt p os se ssi on sti ll a llo ws e rr or s. My discussion of the Twin Ear th thought experiment will make c lear er a nd put to use much of wha t I have sa id about thought e x periments, who should par ticipate in them, the

PAGE 155

146 conditions for conc ept possession, and the sourc es of e rror and cor rec tion in thought experiments. Furthe rmore, the varia tion of Putnam's Twin Ea rth that I prese nt will serve as a pa rallel c ase to the zombie thoug ht experiment that I analy ze in Chapter 5. P ut nam 's T win Ear th Thou ght Expe r im e nt I n this section, I will evaluate Putnam' s Twin Earth thoug ht experiment and a varia tion of it in order to put to use some of the a ppara tus I have a lrea dy introduced (Putnam 1975). I n doing this, I will also be lay ing the g roundwork f or my analy sis of the zomb ie tho ug ht e xpe ri me nt i n Ch a pte r 5 of thi s d iss e rt a tio n. I ha ve a lr e a dy do ne so me backg round work f or a discussion of the Twin Ear th thought experiment in Chapter 1 of the dissertation, whe re I introduce the standard a ccount of a posterior i nece ssities and sk e tc h th e be g inn ing s o f a qu a sitr a dit ion a l th e or y of c on c e pts . T he ma in p ur po se of thi s section is to expl ain how we can f all into error in doing a thought e x periment e ven whe n we ha ve prope rly perf ormed it, that is, to ex plain how a thoug ht experiment can be pr op e rl y pe rf or me d, bu t no t su c c e ssf ull y pe rf or me d. An oth e r p ur po se of the c ha pte r i s to show how the Twin Ea rth thoug ht experiment is supposed to show that the traditional theory of mea ning is mistaken, a nd, thus, how traditional theories of conce pt may be mistaken. Twin Ea r th Thou ght Expe r im e nt Th e fo llo wi ng is t he Tw in E a rt h th ou g ht e xpe ri me nt a s Pu tna m pr e se nts it: [S]upp os e tha t so me wh e re in t he g a la xy , th e re is a pla ne t w e sh a ll c a ll T wi n Earth. Tw in Earth is very much like Ear th; in fact, people on Twin Ear th even speak En gli sh . I n fac t, apar t from the diffe renc es we shall specify in our scienc efi c tio n e xamp le s, [y ou ] ma y su pp os e tha t T wi n E a rt h is exac tly like Earth. [You] may even suppose that he has a Doppelgänger —a n id e nti c a l c op y —o n T wi n Earth . . . Although some of the pe ople on Twin Ear th (say , the ones who c all themselves ' Am e ri c a ns ' a nd the on e s w ho c a ll t he mse lve s ' Ca na dia ns ' a nd the on e s w ho c a ll

PAGE 156

147 ' Di a le c t' ma y be the wr on g wo rd to u se he re , s inc e Put na m sh ou ld n ot w a nt i t to 37 s e e m t h a t o n e w o u l d n o t i c e b y w a t c h i n g p e o p l e t h a t t h e y a r e u s i n g w o r d s d i f f e r e n t l y. Kripke pr esents a similar thoug ht experiment about g old. See Kripke 1980, 38 124-125. themselves ' Eng lishmen,' e tc.) spea k Eng lish, there ar e, not surprising ly , a fe w tin y dif fe re nc e s w hic h w e wi ll n ow de sc ri be be tw e e n th e dia le c ts o f E ng lis h spoken on Twin Ea rth and Standar d Eng lish. . . . 37 2 On e of the pe c uli a ri tie s o f T wi n E a rt h is tha t th e liq uid c a lle d ' wa te r' is n ot H O but a diffe rent liquid whose che mical formula is ve ry long a nd complicated. I sh a ll a bb re via te thi s c he mic a l f or mul a sim ply a s X YZ . I sh a ll s up po se tha t X YZ is indisti ng uishable fr om water at normal tempe rature s and pre ssures. I n particula r, it tastes like water and it quenche s thirst li ke wa ter. Also, I shall suppose that the ocea ns and lake s and sea s of Twin Ear th contain X YZ and not wate r, that it rains X YZ on Twin Ear th and not water , etc. (Putnam 1975, 223) Ca noni c al q ue st ion : I s there w ater on Twin Ear th? 38 I n this presentation of the thoug ht experiment, we ar e told the che mical composition of what Twin Ea rthlings c all ' water ' on Twin Ear th. Of cour se g iven this description of the 2 targ et sce nario, we can inf er tha t water is H O. Howe ver, w e nee d be able to make that infere nce—w e just need to know tha t the clea r, drinkable liquid that fill s lakes and r ivers 2 is actually H O (that is, we ne ed to be in a c erta in epistemic position with respect to the a c tua l w or ld) , to g e t th e ri g ht r e su lt o f t he tho ug ht e xpe ri me nt. We c a n b e in t his 2 epistemic position whether w e know that wa ter is H O or we are g iven stipulated 2 backg round informa tion stating that in the ac tual world wa ter is H O. Put na m ( 19 75 ) a rg ue s, I be lie ve c or re c tly , th a t th e pr op e r r e sp on se to t his c a no nic a l qu e sti on is, " No , th e re is n o w a te r o n T wi n E a rt h. " Th e re is n o w a te r o n T wi n 2 Ea rt h b e c a us e ne c e ssa ri ly , w a te r i s H O, a nd the stu ff on Tw in E a rt h, tho ug h it sh a re s a ll of its superfic ial proper ties with water, is made up of XYZ . Since both the Ear thling and the Tw in E a rt hli ng ha ve the sa me ps y c ho log ic a l st a te s w ith re sp e c t to wh a t th e y c a ll

PAGE 157

148 ' water ' , but there is no wa ter on Twin Ea rth, there must be something more to the meaning of a w ord than wha t is in t he hea d of a spe aker . That is, the traditional theory of meaning is mist aken. Putnam arg ues that there are two assumptions that the traditional theory of me a nin g re sts on , o ne of wh ic h w e mus t g ive up a s a re su lt o f w ha t th e Tw in E a rt h tho ug ht e xpe ri me nt t e lls us a bo ut c e rt a in t e rm s: (I ) T ha t kn ow ing the me a nin g of a te rm is j us t a ma tte r o f b e ing in a c e rt a in psy cholog ical state ( in the sense of ' psy cholog ical state,' in which states of memory and psy cholog ical dispositions are ' psy cholog ical states' ; no one thoug ht that knowing the meaning of a w ord wa s a continuous state of c onsciousness, of course ). (I I ) T ha t th e me a nin g of a te rm (i n th e se ns e of ' int e ns ion ' ) d e te rm ine s it s extension (in the sense that samene ss of intension entails sameness of extension). (Putnam 1975 , 219) Be fore moving on, I will ex plain briefly how Putnam's te rminology here maps onto the te rm ino log y I int ro du c e d in Cha pte r 1 . I n Ch a pte r 1 , I ta lk a bo ut c on c e pts , th ou g hts , predic ates, a nd meaning s (among other thing s). Putnam, of course , only talks about terms and mea nings, but we can e asily chang e his talk about ' water ' to talk about the pr e dic a te ' is w a te r' , a nd ta lk a bo ut t he me a nin g of the pr e dic a te , ' is w a te r' . T he pr e dic a te 2 ' is water' is only true of stuff that is H O. I understand Putnam' s use of ' intension' a s being analog ous to Fre g e' s use of the te rm ' sense' . I n (I ) as we ll as (I I ), Putnam uses ' me a nin g ' to t a lk a bo ut i nte ns ion , b ut i n h is v ie w, the me a nin g of a te rm inc lud e s b oth its int e ns ion a nd e xten sio n. Th e se ns e of a pr e dic a te a c c or din g to F re g e is w ha t I c a ll a ' conce pt', w hile the sense of a pre dicate ma ps onto Put nam' s notion of intension. Howeve r, since F reg e is the one w ho introduced w hat has be come the tr aditional theory of mea ning, he arg ues that the sense of an e x pression dete rmines its refe renc e, or extension.

PAGE 158

149 Acc ording to the traditional theory of mea ning, the re a re two c omponents of the meaning of a w ord, and the y are closely rela ted to one anothe r: (a) there is the intension of the wor d, which is what we ordinarily think of as the mea ning of the term (tha t is, the c on c e pt t he wo rd e xpre sse s) , a nd wh ic h h a s n oth ing to d o w ith the wa y thi ng s tu rn ou t in the world outside one' s head; a nd (b) ther e is the extension or ref ere nt of the term. Although it is often the c ase tha t when two wor ds have the same intension they have the same e x tension, the Twin Ear th thought experiment shows that the tra ditional theory of me a nin g is m ist a ke n in its c la im t ha t th e int e ns ion of a te rm determine s its ex tension. Putnam (1975) nee ds to arg ue that the tra ditional theory makes a third assumption, namely , that we know the intension of ' water ' and this knowing is narrow (not depe ndent at all on the external e nvironment), if his objection is g oing to wor k. For if we do not know the intension of ' water ' , then we do not ha ve the na rrow psy cholog ical states require d to ge t to the claim that two people who ha ve the sa me nar row psy cholog ical states with respe ct to ' water ' can r efe r to diffe rent thing s. That is, we c ould not say that there are two terms with the same intension that diffe r with respe ct to their intension. Assuming that Putnam (1975) a g ree s with the third assumption of the traditional theory of mea ning, the re a re c erta in terms that have the same intension, but which diffe r in ext ension (e.g ., ' water ' uttered on Ea rth and ' water ' uttered on Tw in Earth). Spea kers on Earth a nd Twin Ear th are the same w ith respec t to their narr ow psy cholog ical states when they use ' water ' , but as a matter of fa ct, they are talking a bout differ ent things. What the speake r is talking a bout (or, ra ther, wha t actual stuff the spea ker is re fer ring to) depends, in this case , on wher e the spe aker is uttering the term. So, the intension of a wo rd do e s n ot d e te rm ine its e xten sio n ( i.e ., we mus t g ive up (I I )) . I n o the r w or ds , if intension determined e x tension and the intension, which is wholly a matter of nar row

PAGE 159

150 ps y c ho log ic a l st a te s, a nd the int e ns ion is w ha t w e g ra sp wh e n w e un de rs ta nd a te rm in na tur a l la ng ua g e , th e n tw o p e op le wh os e na rr ow ps y c ho log ic a l st a te s a re the sa me wo uld be using terms that have the same e x tension. But, the Tw in Earth thoug ht experiment shows us that two individuals can be in the sa me nar row psy cholog ical states with re spect to 'w ater ' but refe r to two diffe rent thing s. Consequently , there are two aspec ts of meaning associate d with any g iven term, one that we a rrive a t a priori, and the other that we ( may ) ar rive a t a posteriori, re sulting in a two-dimensiona l semantics. I discussed two-dimensional semantics in the se ction of the pre vious chapter entitled, "I ntensions and Two-Dimensiona l Semantics." Va r iat ion of t he Twin Ea r th Thou ght Expe r im e nt Th e fo llo wi ng is a va ri a tio n o f t he Tw in E a rt h th ou g ht e xpe ri me nt, wh ic h, I wi ll a rg ue , is a na log ou s to the zomb ie tho ug ht e xpe ri me nt I dis c us s in Cha pte r 5 of thi s dissertation: Suppose that somewher e in the g alaxy , there is a planet we will call ' Twin Ear th'. Twin Ear th is very simil ar to Ea rth: people on Twin Ea rth speak a lang uag e that sounds exactly like Eng lish, and for e ach pe rson on Ear th there is a pe rfe ct phy sical duplicate of him or her on Twin Ear th. What Earthlings c all ' water ' on Ea rt h a nd wh a t T wi n E a rt hli ng s c a ll ' wa te r' on Tw in E a rt h s ha re the sa me superf icial qualities: what Twin Ea rthlings c all ' water ' on Twin Ear th looks and ta ste s ju st a s w ha t E a rt hli ng s c a ll ' wa te r' on Ea rt h, wh a t T wi n E a rt hli ng s c a ll ' wa te r' on Tw in E a rt h q ue nc he s th ir st j us t a s w ha t E a rt hli ng s c a ll ' wa te r' on Ea rt h do e s, wh a t T wi n E a rt hli ng s c a ll ' wa te r' on Tw in E a rt h f ill s la ke s a nd oc e a ns jus t as wha t Earthling s call ' water ' on Earth doe s, ' water ' is what Twin Ear thlings say comes out of the sky when it is raining , and so on. Ca noni c al q ue st ion : I s there w ater on Twin Ear th? The a nswer tha t one g ives to the ca nonical question may depend a t least in part on what fur ther infor mation one has a bout the actua l world, that is, the context one is in, and, per haps, informa tion about Twin Earth. Suppose that the pa rticipant has no f urther inf or ma tio n a bo ut t he a c tua l w or ld t ha t mi g ht b e re le va nt t o th e tho ug ht e xpe ri me nt;

PAGE 160

151 su pp os e , f or ins ta nc e , th a t it is 1 75 0, a nd no on e kn ow s th a t w ha t E a rt hli ng s c a ll ' wa te r' is 2 H O bec ause it is prior to the deve lopment of atomic theory . I t seems that many participa nts will say , "Y es, there is water on T win Earth." After all, what we call ' water ' and wha t Twin Earthling s call ' water ' share all of their supe rficia l properties a nd function in t he sa me wa y . A lth ou g h th e re ma y be a na tur a l te nd e nc y to a ns we r t he qu e sti on thi s way , it is not the cor rec t refle ctive re sponse of someone who is pay ing a ttention to what is g oin g on . T he pr op e r r e sp on se to t his c a no nic a l qu e sti on is " I do n' t kn ow ," or " I t is epistemically indeterminate ," be cause we do not know whe ther the liquid ca lled ' water ' on Earth a nd the liquid called ' water ' on Twin Ear th have the sa me ultimate composition. When provided with furthe r bac kg round informa tion, such as that the wa tery stuff on 2 Ea rt h is H O a nd the wa te ry stu ff on Tw in E a rt h is XY Z , e pis te mic ind e te rm ina c y sh ou ld be eliminated, a nd the par ticipant should say that there is no water on T win Earth. One may arg ue that the r eason w e have the intuition we do in this thought experiment is that we do not fully possess the conc ept of water . Howeve r, there is a way of under standing how we c an fully possess a natur al kind conce pt without having the re le va nt b a c kg ro un d k no wl e dg e . T he a bil iti e s th a t on e wh o p os se sse s a c on c e pt m us t have c an be r epre sented by means of a series of conditionals, particular ly in case s where we do no t kn ow a bo ut t ho se fe a tur e s o f t he a c tua l w or ld t he c on c e pt p ic ks ou t. T ha t is , some of the simple conc eptually true propositions involving the c oncept that one must be able to re cog nize may be in the for m of a c onditional. I introduced c onditional conce ptual knowledg e the se ction of Chapter 2 entitled, "Conc eptual Truth, Conce ptual Kn o wl ed ge, an d C o n ce p t u al An al y s i s . " T h i s k i n d o f c o n d i t i o n al co n ce p t u al k n o wl ed ge is releva nt to natural kind conc epts, which a re disting uished by the fa ct that which proper ty they pick out depends on f eatur es of the a ctual conte x t. Even if we believe w e

PAGE 161

152 This impli cit knowledg e is a kind of knowhow, and so we can f ail to draw the 39 ri g ht i nf e re nc e s f ro m th e c on dit ion a ls. k n o w t h e f e a t u r e s o f t h e a c t u a l w o r l d t h a t d e t e r m i n e w h i c h p r o p e r t y i s p i c k e d o u t b y a c on c e pt, we sti ll h a ve the c on dit ion a l kn ow le dg e a s w e kn ow wh a t to sa y if we a re in e rr or a bo ut t ho se fe a tur e s. Sup po se tha t a pa rt ic ipa nt i n a tho ug ht e xpe ri me nt w ho po sse sse s th e c on c e pt i n question, (say , N ) knows a pr iori the following conditionals (wher e ' B' is backg round information, ' N ' is a natura l kind concept, a nd ' C' is a c hemical pr operty ): (1) I f B , then N is C . (2) I f B *, then N is n ot C . On e ma y wo rr y sti ll t ha t ha vin g thi s so rt of kn ow le dg e is n ot s uf fi c ie nt f or fu lly possessing the conce pt. On the contra ry , we ne ed to make f ull concept possession su ff ic ie ntl y thi n s o th a t pe op le wh o k ne w n oth ing of the na tur e of wa te r c ou ld b e sa id t o fully possess the conc ept of water . Fur ther, the a ccount of full conce pt possession given e a rl ie r f its thi s a c c ou nt q uit e nic e ly . T his kin d o f c on dit ion a l kn ow le dg e (i n a dd iti on to other knowle dg e that one ha s about the conc ept), in ca ses wher e we know the re levant fea tures of the actua l world, g ives the par ticipant enoug h to be ab le to say acc urate ly and with justi fica tion whether a conce pt applies to a re levantly completely descr ibed targ et sc e na ri o. Th is a bil ity ne e d n ot a lw a y s b e e xer c ise d, a s w he n o ne fa ils to b e se ns iti ve to backg round informa tion. I n case s where one does not know the rele vant bac kg round informa tion, for example, cer tain case s involving natura l kind concepts, thing s are a bit more c omp lic a te d. I n th os e c a se s, on e wi ll o nly kn ow the se c on dit ion a ls i mpl ic itl y , a nd , a s a 39 r e s u l t , m a y f a i l t o r e c o g n i z e t h e i r r e l e v a n c e t o a p a r t i c u l a r t a r g e t s c e n a r i o . T h a t i s t o s a y,

PAGE 162

153 since the pa rticipant does not know that B is true or B * is true, and so for th, in the actual wo rl d, e ve n th ou g h s he kn ow s how things would be if B w ere the ca se or B * wer e the ca s e, an d s o fo rt h , i n t h e a ct u al wo rl d , s h e m ay j u s t n o t re al i z e, wh en an s we ri n g a canonic al question, that bac kg round informa tion is relevant. As a result, her intuiti ons do no t se rv e a s g oo d e vid e nc e fo r a ph ilo so ph ic a l po sit ion , b ut, a g a in, sh e c a n s til l po sse ss the re levant conc ept. This seems likely to happen e specia lly when one has not had he r attention draw n to differ ent conse quence s that differ ent bac kg round informa tion may have f or an a nswer to a canonic al question. I f a pa rticipant is only g iven one ve rsion of a thought e x periment without any stipulated backg round informa tion and knows a bunc h of these c onditionals, but not which, if a ny , is actually true a bout the world, then she ma y jus t no t be se ns iti ve to t he re le va nc e of the m. F ur the rm or e , s he ma y ma ke po ssi bly fa lse backg round assumptions about the conc ept in question, which may be broug ht to light by prese nting thoug ht experiments that vary on the par ameter of bac kg round informa tion. The kind of c onditional knowledg e I descr ibed above makes sig nificant room f or err or in thoug ht experiments, especia lly if the par ticipant is not provided with stipul ated ba c kg ro un d in fo rm a tio n. Ha vin g sti pu la te d b a c kg ro un d in fo rm a tio n ma y e lim ina te po ssi bly fa lse ba c kg ro un d a ssu mpt ion s th a t th e pa rt ic ipa nt h a s a bo ut t he c on c e pt i n qu e sti on , w hic h ma y inf lue nc e he r r e sp on se to t he c a no nic a l qu e sti on . T his wi ll b e c ome muc h c le a re r w he n I de sc ri be wh a t I thi nk is g oin g on in t he zomb ie tho ug ht e xpe ri me nt.

PAGE 163

154 CH APT ER 4 H IS T O R IC A L C H A LLE N G E S T O C O N C E P T U A L A N A LY S IS Any defe nse of c onceptua l analy sis and thought e x periments should include a se c tio n o n s e ri ou s c ha lle ng e s th a t ha ve be e n r a ise d a g a ins t c on c e ptu a l a na ly sis . I n th is chapte r, I will address two major c halleng es to conc eptual ana ly sis, the para dox es of a na ly sis a nd Qu ine ' s a tta c ks on c on c e ptu a l a na ly sis . T he tw o ma in p a ra do xes I wi ll dis c us s, Me no ' s p a ra do x in Pla to' s Meno (1997) a nd Moore' s (1942) skepticism about conce ptual analy sis, are one s acc ording to which there is a problem about the na ture of c on c e ptu a l a na ly sis —in or de r t o k no w w he the r w e ha ve re a c he d a n a na ly sis of a c on c e pt, we a lrea dy have to ha ve an a naly sis of it, so that any conce ptual analy sis is going to be either trivial or incorre ct (Plato 1997). So, it does not look as if we c an lea rn any thing new by doing c onceptua l analy sis. I will also discuss Quine' s (1960, 1980) re jections of the ana ly tic-sy nthetic distinction and the notion of meaning and explain how they are rela ted to skepticism about conce ptual analy sis. I arg ue that we can use a twodim e ns ion a l se ma nti c s to a lle via te so me of the sk e pti c ism . P ar ado x o f Ana lys is I n this section, I will discuss Meno's pa radox, its relation to the paradox of a na ly sis , a nd Mo or e ' s p os iti on a bo ut t he pa ra do x of a na ly sis (M oo re 19 42 ). B oth pa ra do xes r a ise sim ila r p ro ble ms a bo ut t he na tur e of c on c e ptu a l a na ly sis —in or de r t o know whethe r we have r eac hed an a naly sis of a conc ept, we a lrea dy have to ha ve an a na ly sis of it, a nd a ny c on c e ptu a l a na ly sis is g oin g to b e e ith e r t ri via l or inc or re c t,

PAGE 164

155 respe ctively . So, it does not l ook as if we can le arn a ny thing ne w by doing c onceptua l analy sis. I f we cannot lea rn any thing ne w by doing c onceptua l analy sis, then it is unclea r whether we should bother doing it at all. I f we should not bother doing c onceptua l analy sis at all, it is unclea r wha t we should be doing , if we still want to do philosophy . Cle a rl y , if we wa nt t o d e fe nd a tr a dit ion a l vi e w a bo ut w ha t w e a re do ing wh e n w e c la im to be doing philosophy , we should take seriously this kind of skepticism and try to show that it is unwarra nted. M e no' s P ar ado x a nd t he P ar ado x o f Ana lys is I n Meno Plato (1997) introduce s Meno' s para dox acc ording to which there is an internal incohe renc e whe n we try to give an informa tive analy sis of a conc ept. Either we alre ady have a n analy sis, in which case we do not nee d to look for one, or we do not have a n a na ly sis , in wh ic h c a se we wi ll n ot k no w i t w he n w e g e t on e . M e no sa y s: Ho w w ill y ou loo k f or it [v ir tue ], So c ra te s, wh e n y ou do no t kn ow a t a ll w ha t it is? How will y ou aim to searc h for something y ou do not know at all? I f y ou should meet with it, how will y ou know that this is the thing that y ou did not know? Soc ra te s r e pli e s: I know what y ou want to say , Meno. Do y ou rea liz e wha t a deba ter' s arg ument y ou are bring ing up, tha t a man ca nnot searc h either f or wha t he knows or f or wh a t he do e s n ot k no w? He c a nn ot s e a rc h f or wh a t he kn ow s— sin c e he kn ow s it , there is no need to sea rch—nor for wha t he does not know, for he does not know what to look for. (Plato, 880) Socrate s does not take Me no' s para dox seriously . I n fac t, it looks as if he doe s not think that the par adox is at all problema tic. He e x plains why it is not a proble m by appea ling to his theory of re collection to explain that it does not m atter tha t we alr eady have knowledg e of a naly ses (a t least in a sense ). Appe aling the theory of re collection may

PAGE 165

156 work for Plato, but thos e who do not ac cept the the ory have to f ind another w ay of e xp la in in g a w a y th e p a r a d o x. I understand Me no' s para dox to present us a t least in part with a dilemma a bout c on c e ptu a l a na ly sis , s inc e wh a t w e do in p hil os op hy is c on c e ptu a l a na ly sis , a nd Pla to talks about conc eptual ana ly sis in many of his other wor ks (Plato 1997). Althoug h we should take Meno' s para dox seriously , it seems that the fra mework f or doing thought experiments that I introduce e arlier in the dissertation ca n help us understa nd why there rea lly is not a dilemma of this kind at all. As I understand it, the dilemma ha s the fo llo wi ng tw o h or ns : e ith e r ( a ) w e kn ow wh a t th e a pp lic a tio n c on dit ion s o f o ur c on c e pts are , in which ca se we do not have to bother looking f or them, or ( b) we do not know our conce pt's a pplication conditions, in which case , we will not know if we have disc overe d them. This presupposes that possessing a conc ept re quires knowing its application conditions. As I arg ue in Chapter 3, one who is a suitable ca ndidate for participa tion in a thought e x periment (w hich, of c ourse, is the primar y method of doing conce ptual a na ly sis ) m us t po sse ss t he c on c e pt t o b e a na ly zed . T his wo rr ie s Me no be c a us e he se e ms to think that possessing a c oncept r equire s much more than I think it does. However , the way I understand c oncept possession lea ves room for us to learn something new by doing conce ptual analy sis. Me no fa ils to d ist ing uis h b e tw e e n p os se ssi ng a c on c e pt a nd kn ow ing e xplic itl y the applica tion conditions of a conce pt. Once w e make the distinction, we should see tha t th e re is n ot q uit e the pr ob le m Me no thi nk s th e re is. Me no wo uld be c or re c t in thinking that ther e is a dilemma if possessing a conc ept re quires knowing its application co n d i t i o n s . W h en we d o co n ce p t u al an al y s i s , we ar e u s u al l y l o o k i n g fo r a co n ce p t 's

PAGE 166

157 application conditions; that does not seem c ontroversia l. But, if possessing a conc ept require s knowing expli citly its application conditions, as Meno see ms to thi nk, then surely we would ha ve to know a c oncept' s application conditions befor e we analy zed it. Successf ul conce ptual analy sis can be done succe ssfully only by someone who posse sses the conc ept in question (and, pe rhaps, some othe rs). What we ha ve to know to do conce ptual analy sis is not what we are looking f or, name ly , the applica tion conditions for ou r c on c e pts . O ne wh o p os se sse s a c on c e pt d oe s n ot h a ve to k no w e xplic itl y its a pp lic a tio n c on dit ion s; p os se ssi ng a c on c e pt, on my a c c ou nt, is h a vin g c e rt a in a bil iti e s: A fu lly po sse sse s c on c e pt C if and only if (a) A is able to rec og nize a priori as true s u ff i ci en t l y s i m p l e c o n ce p t u al l y t ru e p ro p o s i t i o n s i n v o l v i n g C a nd oth e r c on c e pts which A fully possesses; and (b) A can sa y acc urate ly (and w ith just ification) about any rele vantly completely descr ibed targ et sce nario whe ther C a pp lie s, fa ils to apply , or it is objectively indeterminate or epistemica lly indeterminate whether C a pp lie s. Requiring one to know applica tion conditions seems to be too strong f or conc ept possession; it would t urn out that we do not possess many conce pts at all. One of the conseque nces of this account of c oncept possession is that one who posse sses a c oncept should, in principle, have the ability to fig ure out its application conditions without having to know its application conditions at the outset. We should know when we ha ve the cor rec t analy sis, then, not because we kne w the applica tion conditions beforeha nd, but beca use we have c erta in abilities with respect to the c oncept in question that we have used cor rec tly . Now, re ading clause (a) may seem to come close to re quiring that the per son know the applica tion conditions of a conce pt to possess it. For we c an imag ine a c oncept that has as a statement of it applica tion conditions a sufficiently simple conce ptual truth. I n such a c ase, posse ssing the c oncept r equire s being able simply to re c og niz e that

PAGE 167

158 Of c ou rs e , th is i s id e a lize d a bit , s inc e we ve ry ra re ly a c tua lly e nd up wi th f ull 1 fl e dg e d a pp lic a tio n c on dit ion s f or ou r c on c e pts . analy sis as true. This might indee d happen in some c ases, but in many other c ases, the analy sis of a conc ept is not one of those ve ry simple conce ptual truths. Analy ses ar e ra the r m or e c omp le x, a nd on e fi g ur e s th e m ou t by e xploi tin g the a bil ity de sc ri be d in clause (b). W e c an re s p o n d t o b o t h h o rn s o f t h e d i l em m a b y d i s t i n gu i s h i n g p o s s es s i n g a conce pt from knowing its application conditions. By distinguishing c oncept possession from knowledg e of a pplication conditions, we ca n explain that we do have to ha ve knowledg e of how to apply our conc epts to analy ze them, but we lear n something ne w when we do conce ptual analy sis, namely , a conc ept' s application conditions. So, the 1 first horn of the dilemma is not problematic. As for the sec ond horn of the dilemma, w e can a rg ue that althoug h we do not know a conce pt's a pplication conditions prior to doing conce ptual analy sis (possessing a conce pt is consistent with such lack of knowle dg e), w e do ha ve so me kn ow le dg e of ho w t o a pp ly the c on c e pt i n q ue sti on a nd we sh ou ld, in principle, be able to fig ure out whe n we ha ve g otten the applica tion conditions right (that is, we should be able to fig ure out whe n we ha ve propositional knowledg e about the conce pt in question). We move from knowing how to apply a conc ept to knowing that the applica tion conditions are suchand-suc h by doing thoug ht experiments and de te rm ini ng wh e the r t he c on c e pt i n q ue sti on a pp lie s, fa ils to a pp ly , o r i t is ind e te ri min a te whether it applies to the targ et sce nario. Af ter using our know-how to ge t to our int uit ion s a bo ut a ns we rs to c a no nic a l qu e sti on s a nd ru lin g ou t a nd a c c e pti ng c e rt a in conditions as application conditions, we should eve ntually acquir e propositional kn ow le dg e of a pp lic a tio n c on dit ion s o f t he c on c e pt.

PAGE 168

159 Meno' s para dox can be taken a s an epistemolog ical objec tion to conceptual analy sis since it cites a proble m with how we come to know analy ses; we e ither alre ady know them, in which ca se we should not bother analy zing our c oncepts, or w e will not kn ow wh e n w e g e t c or re c t a na ly se s. Th e pa ra do x of a na ly sis , th ou g h th e sa me in s pir it, cannot be taken a s an objec tion to conceptual a naly sis on epistemologica l gr ounds, so a response about conc ept possession is not appropriate for dissolving the para dox . Ac c or din g to t he pa ra do x of a na ly sis , a ny thi ng of the fo rm ( œ x)( x is P if f . . . ) i s either trivial, since what is on the left of the biconditional and wha t is on the right of the biconditional have the sa me meaning , or incorr ect, since what is on the left of the bic on dit ion a l a nd the ri g ht o f t he bic on dit ion a l sh ou ld have the same mea ning if the bic on dit ion a l r e fl e c ts a n a na ly sis . A n a pp ro pr ia te re sp on se to t his pa ra do x wou ld b e to explain how what is on the left side of the bic onditional and what is on the rig ht side of the biconditional have the same mea ning in some se nse that does not make the biconditional trivial. G .E. M oo r e on t he P ar ado x o f Ana lys is Gi ve n th a t Mo or e a c tua lly a tte mpt s to g ive a na ly se s o f m a ny c on c e pts in h is wo rk , h e c le a rl y do e s n ot t a ke the pa ra do x of a na ly sis too se ri ou sly . H ow e ve r, he sh ou ld ha ve so me thi ng to s a y a bo ut i t, e sp e c ia lly sin c e he thi nk s th a t th e re a re so me c on c e pts , for e x ample the c oncept of good , which we cannot a naly ze. Although he does not think he has a solution t o the par adox of analy sis, in a reply to C.H. L ang ford ( 1942), he of fer s thr e e ne c e ssa ry c on dit ion s f or g ivi ng a n a na ly sis : I f y ou are to "g ive an a naly sis" of a g iven conce pt , which is the analysandum , y ou must mention, as y our analysans , a conce pt such that (a ) nobody can know that the analysandum applies to an objec t without knowing that the analysans applies to it, (b) nobody can ve rify that the analysandum applies to an objec t without verify ing tha t the analysans applies, (c ) any expression which expresses the

PAGE 169

160 analysandum must be sy nony mous with any expression which expresses the analysans . (Moore 1942, 663) Th e se thr e e c on dit ion s d o n ot prima facie g ive us what I stated is neede d to solve the para dox , since (c ) see ms to ge nera te the par adox, as I explained above. How ever , g iven what Moore say s later in his response to L ang ford, we can r ead ( c) in a way that does not g ener ate the pa radox of analy sis. Consi der the following analy sis and explanation of wh y the re do e s n ot s e e m to be a pr ob le m f or c on c e ptu a l a na ly sis : "T o be a br other is the same thing a s to be a male sibling." The pa radox arises from the f act that, if thi s st a te me nt i s tr ue , th e n it se e ms a s if it m us t be the c a se that y ou would be making exactly the same state ment if y ou said: "To be a brother is the sa me thing a s to be a brothe r." But it is obvious that these two statements ar e not the same, a nd obvious also that nobody would say that by asser ting " To be a brother is to be a brother" y ou wer e g iving a n analy sis of the con cept "bro th er." (Moore 1942, 665) Something more is convey ed whe n we sa y ' To be a brother is the sa me thing a s to be a male sibling' than when w e say ' To be a brother is the sa me thing a s to be a brothe r' . Moore thinks that, intuiti vely , an ana ly sis of one expression must be a diffe rent expression than what is being analy zed, and it must di ffe r in complexit y . The a naly sans has to make e x plicit something that is not expl icit in the analy zed term. Ag ain, Moore do e s n ot c la im t o h a ve a so lut ion to t he pa ra do x of a na ly sis , b ut h e do e s n ot t hin k it is a pr ob le m on int uit ive g ro un ds . Moore c an use the following explanation of what is g oing on in the a bove example. 'I s a brother ' and ' is a male sibling' are nece ssarily coextensive as a ma tter of meaning in Eng lish. But they are not in the strictest sense sy nony mous beca use the second ha s a compositional structure that the first does not, and, he nce, invoke s rules for un de rs ta nd ing e xpre ssi on s w hic h th e fi rs t do e s n ot. F ur the r, the re is a n e pis te mic differ ence betwee n the two—one c ould know expli citly that something is a br other without knowing e x plicitly that it is a male sibling, and vic e ver sa. None theless, the

PAGE 170

161 la tte r e xpre sse s th e sa me c on c e pt a s th e fo rm e r, be c a us e the c omp e te nc e we ha ve in deploy ing the conce pt of a brother is key ed to the prope rty of being male and of being the of fs pr ing of the sa me pa re nts a s a no the r. Ho we ve r, thi s f a c t w hic h is e xpre sse d in our compe tence is not gua rante ed to be known e x plicitly by us in virtue of our ha ving the co m p et en ce . T h e a n al y s i s , 't o b e a b ro t h er i s t h e s am e t h i n g as t o b e a m al e s i b l i n g', makes e x plicit, then, what is impli cit in our compete nce in the use of ' is a brother ' or competenc e in the deploy ment of the c oncept of a brother . I t is in t his tha t it s u se fu lne ss and informa tiveness lies. I t informs by making expli cit some structure in the conce pt of a br oth e r tha t is e mbo die d in ou r d isp os iti on s to de plo y it. Although Me no' s para dox and the pa radox of analy sis seem at first g lance to be problematic f or conc eptual ana ly sis, after some considera tion, they do not seem pr ob le ma tic a t a ll. I t lo ok s a s th ou g h th e re a re str a ig htf or wa rd re sp on se s to bo th p a r a d o xe s . T h u s , w e c a n le a r n s o me th in g n e w b y d o in g c o n c e p tu a l a n a ly s is . I n th e n e xt chapte r, I will address Quine ' s (1960, 1980) c halleng es to conc eptual ana ly sis, which have be en take n much more se riously than the par adoxes of analy sis by contempora ry philosophers. Quine an A tt ac ks o n Co nc e pt ual Ana lys is Many philosophers who ar e skeptica l about conce ptual analy sis are ske ptical about it on the same g rounds that Quine is skeptica l about it. Most phi losophers u n d er s t an d an al y t i ci t y as t h e f o l l o wi n g: (A1) A sentenc e, S , is analy tic iff S i s t ru e s o l el y i n v i rt u e o f i t s m ea n i n g. There are two kinds of statements that ca n be ana ly tic, statements that ar e log ical truths and ce rtain statements that a re not log ical truths (e .g ., ' All unmarried ma les are unmarrie d' , and ' All bache lors are unmarrie d' , respe ctively ). On Quine ' s view, a log ical

PAGE 171

162 truth is a statement that is true unde r any interpre tation of its terms, excepting log ical constants. Quine is intere sted in those statements that ar e said to be a naly tic, but which are not logica l truths, that is, st atements whic h "c an be tur ned into a log ical truth by pu tt in g sy no ny ms for sy no ny ms " (Quine 1980, 23) . For example, consider the sta tement ' All bache lors are unmarrie d' . One may arg ue that this is analy tic beca use we can substitut e ' unmarrie d males' for ' bache lors' to ge t the resulting statement: ' All unmarried males ar e unmar ried' , which is a log ical truth. I n W ord and Object and " Two Dog mas of Empiricism," Quine ar g ues that there are no analy tic truths on the g rounds that either there is no intelli g ible distinction betwee n analy tic and sy nthetic truths or, insofar as we can ma ke sense of it, there a re arg uments to show the extension of 'is ana ly tic' is empty , and, thus, all truths are on a par ; no sp e c ia l se t of ' a na ly tic ' on e s c a n b e se t a sid e a nd tr e a te d d if fe re ntl y fr om t he re st (Quine 1960, 1980) . One who ha s a Quinea n view about a naly ticity is invariably g oing to be skeptica l about conce ptual analy sis since the re sults of conce ptual analy sis are supposed to be conc eptual truths, which a re e x presse d by analy tic truths. There are sever al ar g uments that Quine g ives ag ainst analy ticity . I n "T wo Dog mas of Empiricism," Quine (1980) arg ues that no statements a re a naly tic on the g rounds that there are four possible wa y s we c an distinguish ana ly tic sentenc es fr om sy nth e tic on e s, e a c h o f w hic h f a ils to s e rv e a s a n e xpla na tio n o f t he dis tin c tio n: a pp e a l to d e f i n i t i o n s , i n t e r c h a n g e a b i l i t y sa lv a v e rit ate , semantica l rules, and ve rifiability in the fac e of a ny possible experience . Since ea ch of the se fa il as an explanation of the

PAGE 172

163 Qu ine ne ve r a rg ue s f or the c la im t ha t th e fo ur de fi nit ion s h e of fe rs a re the on ly 2 possible ones that ca n be g iven. I t is unclear w hy he does not think there are more de fi nit ion s to tr y ou t. distinction, and they are exhaustive, we ha ve no g ood rea son to suppose there is such a 2 distinction. Thematica lly the pape r fa lls int o two parts. I n the first, comprising the first three su g g e sti on s f or ma kin g se ns e of the a na ly tic /sy nth e tic dis tin c tio n, Qu ine a rg ue s th a t a ll of these w ay s of try ing to make sense of it invoke notions t hat ar e as suspe ct as that of analy ticity , viz., t hose of mea ning, sy nony my , and nec essity . Thus, he ar g ues that all of the se a tte mpt s to e xpla in a na ly tic ity a re un su c c e ssf ul. Ab ou t th is p a rt of the pa pe r I wi ll not say much. The a rg ument rests on the a ssumption t hat we c an g ive a r eductive analy sis of analy ticity , but that require s g iving a reduc tive analy sis of meaning , and ther e is no reason to think that this can be done or that it must be done to make se nse of the family of notions of which it is a par t. And, clea rly , most contemporar y philosophers do not reg ard the se notions as in need f or some re duction to be conc epts in g ood standing . I n a dd iti on , to un de rt a ke su c h a g e ne ra l de fe ns e of the se no tio ns wo uld be a big ta sk a ll be itself. Given tha t ge nera lly philosophers are quite comforta ble with them, we ne ed not underta ke that task he re. I n th e se c on d th e ma tic pa rt , Q uin e c on sid e rs wh e the r a pp e a l to ve ri fi c a tio nis m will help. Here his targ et is the L og ical Positivi sts. The L og ical Positivi sts were committed to verifica tionism , acc ording to which the mea ning of a stateme nt is to be so ug ht i n it s me tho d o f v e ri fi c a tio n, a nd no sta te me nt i s me a nin g fu l e xce pt i ns of a r a s it is verifia ble (or a naly tic or contra dictory , but these ca n in turn be def ined in terms of

PAGE 173

164 I n Q uin e (1 96 0) , h e a rg ue s th a t th e be st w e c a n g e t f ro m a ny no tio n o f a na ly tic ity 3 is the notion of stim ulus analy ticity (SA): (SA) A se ntence , S , is sti mul us a na ly tic jus t in c a se a sp e a ke r h old s S true no matter what , that is, the speake r is prepa red to a ssent to S "com e wh at m ay ." This however is after he re jects (A2) , and is by way of a r econstruc tion of the notion of analy ticity in behavioristically respe ctable te rms. Th e ve ri fi c a tio nis t pr inc ipl e se e ms t o b e se lf -d e fe a tin g (u nle ss i t is a na ly tic )— if 4 the on ly wa y a se nte nc e c a n b e me a nin g fu l is fo r t he re to b e a me tho d f or ve ri fy ing it, then the ver ification principle itself ne eds to be ve rified. B ut, it does not seem that there is a ny e xpe ri e nc e of ou rs tha t c a n v e ri fy it. c on fi rm a bil ity ). Th e de fi nit ion tha t th e Pos iti vis ts w ou ld o ff e r i n te rm s o f c on fi rm a bil ity would be (A 2). (A2) A sentenc e, S , is analy tic iff S is confirmed no ma tter wha t experience we can ha ve. 3 Why is (A2) the na tural wa y to define a naly ticity for the ve rifica tionist ? (A2), of course , is consistent with a principle of ve rifica tion: a sentence that is analy tic is verified in the fac e of a ny possible experience . But in lig ht of the idea that meaning is a method of ve ri fi c a tio n, if we thi nk a sta te me nt h a s a me tho d o f v e ri fi c a tio n w hic h g ua ra nte e s it is alway s true, then it is true in virtue of its method of ve rifica tion, i.e., meaning , and, hence , is analy tic. (A2) is a lso consistent with Quine' s insist ence that fac ts about 4 meaning are inhere ntly rela ted to verba l behavior ( Quine 1960, 1980). Since we un de rs ta nd the me a nin g of a sta te me nt i n te rm s o f i ts m e tho d o f v e ri fi c a tio n a nd a ll f a c ts a bo ut m e a nin g mus t be pu bli c ly a va ila ble , a me tho d o f v e ri fi c a tio n o f a sta te me nt m us t be pu bli c ly a va ila ble . I n tr y ing to f ig ur e ou t w ha t a sp e a ke r' s w or ds me a n, we ou g ht t o rely on observa tion of the speake r' s behavior, in pa rticular, on he r re sponses to various stimul i in her environme nt, rather than any thing e lse.

PAGE 174

165 Quine wor ried that ther e ar e sente nces w e may have thoug ht were analy tic, but which ar e ra tionally revisable . The two-dimensiona l picture he lps, since the sec ondary intension of a ter m solely depends on how the ac tual world turns out. I f the sec ondary intension of a stateme nt we thoug ht was ana ly tic differ s from its primary intension, then we should be w illing to chang e our minds about whe ther it is analy tic g iven furthe r experience . What is actually g oing on w hen we have to c hang e our hold-tr ue attitude toward a sentenc e we thought wa s analy tic is that we have discovere d its secondar y intension. For e x ample, it is plausible to think of the following, or something similar, a s a pp ro a c hin g the de fi nit ion of ' g old ' : G old is a ma lle a ble , d uc til e y e llo w m e ta l. I n f a c t, people may very well have seen this as a g ood definition of ' g old', or as an a naly tic truth, but we no long er do. O n Quine' s story , they no longe r do bec ause e mpirical pre ssures drove us to re ject wha t was appa rently an ana ly tic truth, which shows that it was not rea lly analy tic. I t may be that empiric al pre ssures drove us to rejec t it and that it was not analy tic, and people may have be en led f alsely to think that it was. B ut if they had the too ls o f t wo -d ime ns ion a l se ma nti c s a t th e ir dis po sa l, t he y wo uld ha ve ha d a n e a sie r t ime seeing what close ly rela ted statement is ana ly tic. There is no need for skepticism about o u r a b i l i t y t o d i s t i n g u i s h s o m e t h i n g t h a t r e a l l y is not rationally revisable in the fac e of new e mpirical evide nce. The two-dimensiona l approa ch sug g ests that there is one sense of ' meaning ' which makes ' g old is a malleable, duc tile y ellow metal' a truth of mea ning—na mely , of the primary intension associated w ith the term. So people found it ea sy to think about the primary intension and used that to help for mulate a de finition. But the sense tha t is at iss ue wh e n w e re je c t ' g old is a ma lle a ble , d uc til e y e llo w m e ta l' a s n ot a g oo d d e fi nit ion is t h e s e c o n d a r y i n t e n s i o n . T h e t w o d i m e n s i o n a l a p p r o a c h a l s o h e l p s i n t h e f o l l o w i n g w a y:

PAGE 175

166 i t p o i n t s t o w h a t r e a l l y is analy tic, namely , the conditional: ' if the ac tual malleable, du c til e y e llo w m e ta l is X, the n g old is i de nti c a l w ith X' . T ha t w a s n e ve r r e je c te d, no r i s it likely to be re jected. Quine may , of cour se, insist that even the c onditional statement that I have a rg ued is analy tic might be g iven up ra tionally in the fac e of ne w evidenc e. The advoca te of analy ticity , howeve r, ca n insist t hat any such g iving up is re ally just a chang e of meaning , a g iving up the se ntence , not of what it said. The possibility of this maneuver looks to put t he burde n of proof back on Q uine. Howeve r, Quine thinks that there is no such a thing as a c hang e of me aning be c a us e the re is n o w a y to d e te rm ine wh e the r s ome on e ha s h a d a c ha ng e of be lie f o r i s chang ing the meaning of her words. Quine a rg ues that there are circ umstances in which one may chang e the mea ning of his terms, but in which we c annot know whe ther he has c ha ng e d th e me a nin g of his te rm s o r h a s c ha ng e d w ha t he thi nk s a bo ut t he wo rl d; a ll translations are underde termined, a nd, Quine infe rs, there fore indeterminate . For example, consider the f ollowing se ntence that people may think is analy tic: (K) I f someone know s that P, then it i s true that P. We find it hard to see how any empirica l evidence can r ender (K) f alse. Quine will ag ree that no single experience would be able to rende r (K) false. H oweve r, he w ill ins ist that ov e ra ll c on sid e ra tio ns c ou ld r e nd e r ( K) fa lse in t he fo llo wi ng wa y . I f w e c on tin ue to hold (K) true , then we ma y be for ced to a dmit that a lot of things people sa y they know, the y do no t, i n f a c t, k no w. Sup po se , f or e xamp le , th a t w e ha ve c on c lud e d th a t mo st statements about ordina ry objects (e .g ., tables, tree s, etc.) a re f alse be cause they do not exis t. I n such a c ase, if w e continue to hold (K ) true, w e have to say that most people do not rea lly know that there are tree s (since the re a re, a s a matter of fac t, no trees) and that

PAGE 176

167 no on e re a lly kn ow s th a t he is l oo kin g a t a tr e e , a nd so on . B ut, tha t se e ms counter intuiti ve, since it would mean that there is very littl e that we do know. I nstead of taking back the thesis that there a re no or dinary objects, we might re ject (K ). Rejec ting (K) c ould allow us to say other thing s we wa nt to say (e.g ., that we know tha t there a re tr e e s) a nd the re by sim pli fy thi ng s. Someone who wa nts to defend the a naly tic status of (K) may respond by say ing tha t w e c a n ju st c ha ng e the me a nin g s o f w or ds if we wa nt. Th a t w ou ld e na ble us to pr ov ide a sim ple r o ve ra ll f ra me wo rk . We c a n ju st u se ' kn ow ' to m e a n s ome thi ng e lse , s o that (K) is no long er a naly tic. That wa y , we do not have to chang e the wa y we think about the wor ld. Quine would say that there is no differe nce be tween c hang ing the way we think about the wor ld and cha ng ing the meaning of one' s terms. 1 Consider the following situation: At t , Emma and Matt believe that (K) is true . 2 At t , (a) Emma and Matt both believe (K) is fa lse; (b) Matt and E mma both rea d and be lie ve a n a rt ic le tha t ta lks a bo ut t he re c e nt d isc ov e ry tha t th e re a re , in fa c t, n o o bje c ts i n the world, a nd, (c) as a r esult, her theor y about knowledg e cha ng es so that she a llows that to k no w t ha t p d oe s n ot r e qu ir e it t o b e tr ue tha t p, wh ile (c ) M a tt s imp ly de c ide s to us e (K ) t o me a n s ome thi ng e lse , w hic h d oe s n ot r e qu ir e the tr uth of ' p' fo r ' x kno ws tha t p' to be tr ue of so me on e (M a tt' s li ng uis tic fr a me wo rk c ha ng e s) . So me thi ng ha s c ha ng e d w ith respe ct to both speake rs, but what has c hang ed is diffe rent: whe re the meaning of (K) re ma ine d th e sa me in E mma ' s d ia le c t ( sh e jus t c ha ng e d h e r m ind a bo ut i ts t ru th) , M a tt has not cha ng ed his mind about any thing of substance. T h er e i s , o f c o u rs e, n o wa y t h at o n e c o u l d o b s er v e w h at h ap p en ed i n E m m a's head a nd what happe ned in Matt' s head, so, it may seem that ther e is no way to verify the dis tin c tio n b e tw e e n w ha t E mma no w t hin ks of (K ) a nd wh a t Ma tt do e s. Th e y bo th

PAGE 177

168 continue to endor se (K) . Of c ourse, they might tell us what they wer e thinking. Q uine thinks we could re interpre t those things so that they would be compa tible with either story . I f so, we w ill not find any observa ble diffe renc e which ve rifies that Emma ha s chang ed her mind about something substantive, a nd Matt has just decided to c hang e wha t he mea ns by his words. As a r esult, for Quine , there simply is no disti nction betwee n ch an ge o f b el i ef an d ch an ge o f m ea n i n g. Qu ine ' s a rg ume nt h e re is s up po se d to sh ow tha t si nc e the re is n o w a y to t e ll wh e the r a c ha ng e in o ne ' s h old -t ru e a tti tud e tow a rd a se nte nc e is g ro un de d in a c ha ng e in belief or a cha ng e in meaning , it is m eaning less to say that Emma cha ng es her mind about the wor ld and Matt cha ng es wha t he means by a sente nce. I a m no t w or ri e d a bo ut t his pa rt ic ula r s ke pti c a l w or ry of Qu ine ' s h e re , s inc e it i s on e wh ic h r e lie s o n th e pr inc ipl e of ve ri fi c a tio n. An d if we re je c t th e pr inc ipl e , it se e ms that Quine' s worry is about whether there is a diffe renc e betwe en a c hang e of me aning a nd a c ha ng e in b e lie f i s a n e pis te mic on e . I n th is c a se we c a n— ve ri fi c a tio nis m a sid e —r ule Qu ine ' s w or ry ir re le va nt. F or jus t be c a us e we c a nn ot know what the dif fe re nc e is b e tw e e n E mma a nd Ma tt i n th e a bo ve e xamp le , d oe s n ot m e a n th a t th e re is no differ ence . Now ther e ar e, per haps, some thing s that we c an say in support of Quine (not of his view, nec essar ily , but of his challeng e to g ive an a ccount of analy ticity ). Even if we do not acc ept his arg ument, there is something that we can g et out of it. I n his discussion of ana ly ticity and sy ntheticity , there are sever al words that a re pe rhaps too fr eely used intercha ng eably (e.g ., ' nece ssary ' , ' a priori' , and ' analy tic' ). Since Quine ' s work on analy ticity , various counte rexamples to the equivalenc e of the se three terms have been introduced. F or example, it seems that there are things whic h may be nec essar y , but

PAGE 178

169 Kripke g ives an a ccount of this in 5 (Kripke 1980). which we cannot know a priori, namely , that g old is a metal. I think we ca n take c are of 5 this with a two-dimensional semantics. F urthermor e, ther e see m to be things whic h are analy tic, but not necessa ry (e.g ., ' all actua l philosophers are philosophers' ). Dra wing these distinctions is im portant in fig uring out what is g oing on w hen we talk about analy ticity . But, dra wing more distinctions and doing more c onceptua l analy sis does not seem to support Quine' s broad ske pticism about conce ptual analy sis.

PAGE 179

I say that the thoug ht experiment puts a constraint on the a naly sis of 1 c on sc iou sn e ss beca use the zombie thought e x periment doe s not give us an ana ly sis of c on sc iou sn e ss (i.e., it does not g ive us nec essar y and/or suff icient conditions for a pply ing it). I do not expect the map to be exhaustive, but I hope that it will lay out the 2 territory in a way that is releva nt to issues about conce ptual analy sis. Som e of the po sit ion s a re dis c us se d in so me fo rm in C ha pte r 4 , b ut I wi ll p oin t ou t w he re the y fi t in the map I offe r in this chapter . 170 CH APT ER 5 CONTEMPORARY CASE S TUDY: THE Z OMBI E THOUG HT EXPERI MENT As I a rg ue thr ou g ho ut t his dis se rt a tio n, tho ug ht e xpe ri me nts pla y a c ru c ia l r ole in doing c onceptua l analy sis. I n addition to putti ng a constra int on the analy sis of the 1 conce pt of c on sc iou sn e ss , Cha lme rs ' s zom bie tho ug ht e xpe ri me nt h a s sp ur re d a de ba te in the philosophy of mind about the role a nd rele vance of conc eptual ana ly sis (Chalmers 19 96 ). Th e a pp a re ntl y a nti -p hy sic a lis t r e su lt o f t he tho ug ht e xpe ri me nt h a s le d s ome ph ilo so ph e rs to l oo k mo re c los e ly a t it to s e e wh y we se e m to g e t a n a nti -p hy sic a lis t re su lt w he n w e ima g ine the ta rg e t sc e na ri o. Som e ph y sic a lis ts h a ve loo ke d to the ro le a nd re le va nc e of c on c e ptu a l a na ly sis in t he ph ilo so ph y of min d f or a n e xpla na tio n. I wi ll use Chalmers' s zom bie thoug ht experiment as a wa y of consider ing va rious response s phy sicalists have to conc eptual ana ly sis in the philos ophy of mind. First, I lay out the zombi e thoug ht experiment so that it fit s within the frame work for thinking about thoug ht e xpe ri me nts tha t I pr ov ide d in Cha pte r 2 of the dis se rt a tio n. Th e n, I fi ll i n in mor e de ta il a map ( Fig ure 4) that I have pr ovided, which lay s out differe nt answer s to and ana ly ses of the ca nonical question aske d of par ticipants in the zom bie thoug ht experiment, and I 2 dis c us s in mor e de ta il t ho se po sit ion s w hic h in vo lve iss ue s a bo ut c on c e ptu a l a na ly sis .

PAGE 180

171 Fig ure 4. A map of r esponses to the zombie thought e x periment

PAGE 181

172 Th e re ha s b e e n mu c h d isc us sio n a bo ut w ha t f or mul a tio n o f p hy sic a lis m be st 3 captur es the phy sicalist view. For my purposes, I only need to g ive a ne cessa ry condition fo r p hy sic a lis m. T he on e I ha ve pr ov ide d is , a s f a r a s I kn ow , u nc on tr ov e rs ia l. I n th e lit e ra tur e on thi s to pic , ma ny thi nk tha t on e c a nn ot b e bo th a ph y sic a lis t (a c c or din g to w hic h e ve ry min ima l ph y sic a l du pli c a te of the wo rl d is a c omp le te du pli c a te ) a nd a de fe nd e r o f c on c e ptu a l a na ly sis , s inc e a de fe nd e r o f c on c e ptu a l a na ly sis 3 se e ms b ou nd to s a y tha t zom bie s a re c on c e iva ble , a nd , h e nc e , p os sib le ; bu t a ph y sic a lis t c a nn ot a llo w t ha t zom bie s a re po ssi ble . I ho pe to m a ke ro om f or a ph y sic a lis t r e sp on se to t he c a no nic a l qu e sti on in t he zomb ie tho ug ht e xpe ri me nt t ha t ha s b e e n la rg e ly undiscussed in the literatur e that would allow one to be a pr oponent of c onceptua l analy sis. I n my acc ount of what is g oing on in the zombie thoug ht experiment, I draw an analog y betwee n the conc ept of pa in a nd na tur a l ki nd c on c e pts su c h a s water a nd e xpla in why many participa nts in the z ombie thought e x periment pe rsist in making a cer tain kind of mistake. As fa r as the la rg er pr oject g oes, I am intere sted in what a c lose look at the zomb ie tho ug ht e xpe ri me nt i llu min a te s a bo ut t he re le va nc e of tho ug ht e xpe ri me nts in g ener al, that is, whether we should deny their sig nificanc e or, if the y are signific ant, how they are signific ant. Se tt ing Up t he Z om bie Thou ght Expe r im e nt I n this section, I will lay out the targ et sce nario of the zombi e thoug ht experiment so that it fits i n with the model of thoug ht experiments I prese nted in Chapter 2. The targ et sce nario will set us up for the canonic al question that will serve a s the starting point for the map of response s to the zom bie thoug ht experiment I cre ated ( Fig ure 4) . I wi ll a lso re sta te the c on dit ion s f or pr op e r p e rf or ma nc e of a tho ug ht e xpe ri me nt t ha t I

PAGE 182

173 B y ' ph y sic a l pr op e rt ie s' I me a n th os e pr op e rt ie s th a t a re ta lke d a bo ut i n c ur re nt 4 phy sics or prope rties that would be in the sa me fa mily but not thought of a s mental pr op e rt ie s. introduced e arlier . Then, I will lay out responses that pe ople have g iven to the ca nonical question asked in the zombie thought e x periment. Tar ge t S c e nar io a nd C ano nic al Q ue st ion of t he Z om bie Thou ght Expe r im e nt Th e ta rg e t sc e na ri o o f t he zomb ie tho ug ht e xpe ri me nt i s p re se nte d in sli g htl y differ ent way s in the literature . Some philosophers have de scribed in their targ et scena rio a zombie world while other s have de scribed a n individual z ombie. Chalmers (1996), f or example, descr ibes an individual that is phy sically and func tionally identical to him (his z ombie-twin), but which lac ks phenomena l experience s and ar g ues that beca use " there is no hidden contradic tion lurking in the de scription," the re is no log ical c on ne c tio n b e tw e e n th e wa y the wo rl d is ph y sic a lly a nd the wa y the wo rl d is p h e n o m e n a l l y (Chalmers 1996, 96) . That is, since we can c ohere ntly imag ine the situation, it is conce ivable that ther e ar e zombies, and, so, zombi es ar e possible and phy sicalism is false. This way of pre senting the thoug ht experiment is supposed to show tha t a pr op os e d c on c e ptu a lly su ff ic ie nt c on dit ion (a ph y sic a l de sc ri pti on of the wo rl d) is not sufficient (f or phenome nal experienc e), or , in other words, tha t a phy sical desc ription of the wor ld is not a complete desc ription of the world. Th e fo llo wi ng is o ne wa y we ma y pr e se nt t he ta rg e t sc e na ri o o f t he zomb ie thought e x periment so that it fits into my fra mework: Suppose that there is a being that is identical to me in all of its relational and nonre la tio na l ph y sic a l pr op e rt ie s. Th e be ing e xhibi ts a ll o f t he sa me be ha vio rs tha t I 4 exhibi t, is functionally identical to me, and is in the same phy sical environme nt tha t I a m in —it s st oma c h g ro wl s a t me a lti me , it is d re sse d in c lot he s th a t ma tc h, it rea cts in the same wa y that I rea ct to the same stimuli insofar as a ny non-

PAGE 183

174 I n Chapter 2 in the se ction entitled, "A Fr amewor k for Thinking About Thoug ht 5 Ex periments," I distinguish differ ent kinds of ca nonical questions that one may ask af ter prese nting a targ et sce nario (e .g ., questions that contain modal terms and que stions that contain counte rfa ctuals). See the se ction in Chapter 3 entitled, " Sources of Error and Corre ction," f or a 6 discussion of eac h of these. int e nti on a l, n on -p sy c ho log ic a l de sc ri pti on of it g oe s, it s e e ms t o r e po rt on its internal states, e tc. Ca noni c al q ue st ion : Mu st t he be ing be c on sc iou s? T h e t a r g e t s c e n a r i o i s s t r a i g h t f o r w a r d e n o u g h a n d s e e m s t o b e r e l e v a n t l y c o m p l e t e . It does not look as if there are any extra confusing details to lead our intuitions astray . The canonic al question is a modal var iation of a fir st-order question. 5 P roper P erform ance Re viewed I n Ch a pte r 3 , I dis c us se d th e c on dit ion s u nd e r w hic h s ome on e is a su ita ble candida te for pa rticipation in a thoug ht experiment and the conditions under w hich a targ et sce nario pre sents a re levantly complete de scription of an imag inary case . When we pe rfor m the zombi e thoug ht experiment, we should make sure these c onditions are met. I also explained the conditions for prope r per formanc e of a thought e x periment. Th os e c on dit ion s a re a s f oll ow s: A t ho ug ht e xpe ri me nt i s properly performe d if f ( a ) t he ta rg e t sc e na ri o is rele vantly completely and ca ref ully descr ibed, (b) the participa nt is a competent speake r of the la ng uag e that the c oncept in question is expressed in, (c) the pa rt ic ipa nt ful ly possesses the c oncept in question, (d) the par ticipant understands he r t a sk a s a pa rt ic ipa nt i n a tho ug ht e xpe ri me nt, a nd (e ) t he e xpe ri me nte r m e e ts all of the c riteria a n ideal par ticipant must meet and has a fac ility for de signing tho ug ht e xpe ri me nts . 6 I n most cases, whe n the conditions for prope r per formanc e ar e met, we g et the rig ht result. Consequently , we c an use the results of the thoug ht experiment as a basis for dr a wi ng me ta ph y sic a l c on c lus ion s. Ho we ve r, so me tim e s w he n w e pr op e rl y pe rf or m a

PAGE 184

175 thought e x periment, we g ive an incor rec t answer to the canonic al question bec ause the re is something about the c oncept or the thoug ht experiment itself that prevents us fr om reliably draw ing me taphy sical conc lusions from its results. That is, we can do e very thing rig ht in perfor ming a thought e x periment a nd still not ge t the rig ht answer ; various other fac tors may run interf ere nce. E arlier , I explained how we may fall into this kind of error in perfor ming Putnam' s Twin Earth thoug ht experiment (Putnam 1975). I will arg ue that we make a similar kind of mistake in the zombie thought experiment. I call prope r perf ormanc e of a thought e x periment, the r esult of which c an be use d as evide nce f or or aga i n s t a p h i l o s o p h i ca l t h eo ry 's u cc es s fu l p er fo rm an ce '. Usually the answe r one g ives to a ca nonical question does not dire ctly g ive us a ns we rs to m e ta ph y sic a l qu e sti on s; w e of te n h a ve to d ra w i nf e re nc e s f ro m th e a ns we rs to g et substantive metaphy sical cla ims. I f there is something about the c oncept or the thought e x periment itself that doe s not allow us to make the rig ht infere nces, the n the thought e x periment is not succe ssfully perf ormed. I n succe ssful perf ormanc e of a thought e x periment the e x perimente r should make sure that an idea l participant c an g ive a re liable answe r to the ca nonical question. Ther e may be many conditions that one might ar g ue interf ere with the succe ssful perf ormanc e of a thought e x periment, but I am on ly c on c e rn e d w ith tw o o f t he m he re : (1 ) Th e c on c e pt i nv olv e d in a tho ug ht e xpe ri me nt o r t he tho ug ht e xpe ri me nt i tse lf is exceptional, or diffe rent in substantive wa y s from others. The re a re a t least three thi ng s th a t on e mig ht s a y a bo ut t he zomb ie tho ug ht e xpe ri me nt t ha t ma ke its results untrustworthy or explain why it cannot be suc cessf ully perf ormed: (a) phenomena l conce pts are r ecog nitional concepts a nd as such ha ve diffe rent possession conditions than other conc epts, (b) w e ar e cog nitively closed to the re le va nt f a c ts o r c on c e pts in t his c a se , a nd (c ) t he me c ha nis m of c on c e ivi ng in t his kind of ca se is unreliable (Hill and McL aug hlin 1999; L oar 1990; McG inn 1998). Be cause the conc ept of c on sc iou sn e ss is special in some wa y , some philosophers arg ue, the r esults we g et from imag ining the zombie thoug ht experiment and others involving the conce pt of c on sc iou sn e ss do no t he lp u s in g ivi ng a n a na ly sis

PAGE 185

176 I discuss the notion of stipulated backg round informa tion in C hapter 3 in the 7 section entitled, " Suitable Candidates and Conce pt Possession." of it. They arg ue that althoug h many thought e x periments dea ling with other c on c e pts a re un pr ob le ma tic , th e zomb ie tho ug ht e xpe ri me nt i s e xce pti on a l in some way , and, so, we should not rely on it to help us with giving an ana ly sis of c on sc iou sn e ss . I call these , along with L evine ( 2001) exceptionalist analy ses of what g oes wrong in the zombi e thoug ht experiment. (2) There is releva nt information about the ac tual world that the pa rticipant does not have, the having of which w ould result in chang ing a n answe r to the ca nonical question. I nformation about the a ctual wor ld does not usually make a differ ence to what we w ill say in response to the c anonica l question in a thought e x periment. That is, nothing a bout the actua l world is usually rele vant to the succ essful perf ormanc e of a thought e x periment. F or some conc epts, howeve r, wha t the actua l world is like is relevant to their a pplication conditions, as in the case of 2 natura l kind concepts. Onc e we know that wate r is actua lly H O, for e x ample, we 2 c a nn ot strongly conce ive of wa ter be ing a ny thing other than H O, wher e being able to strong ly conce ive of something means we can imag ine that, g iven the wa y things a re in the a ctual wor ld, they could have been othe rwise. F or example, we 2 cannot strong ly conce ive of XY Z being water beca use nec essar ily , H O is water . I f we cannot strong ly conce ive of XY Z being water , then XYZ is not water. Howeve r, we can w eakly conce ive of wa ter be ing X YZ beca use we can c onceive 2 of us be ing wr on g tha t w a te r i s H O. Som e thi ng is w e a kly c on c e iva ble jus t in case , for a ll we know, we could be mistaken a bout what the ac tual world is like. I f we want re liable re sults from thought experiments involving na tural kind conce pts, then the participa nt must have the re levant empiric al information. Otherw ise, she may not conce ive of the r ight thing . I call these kinds of analy ses of the zombie thought e x periment one s that involve issues about a posterior i n e c e s s i t y. We do not need to include this extra information in the desc ription of the targ et sce nario; rathe r, the e x perimente r should add it to the thought e x periment a s stipulated backg round information. Sometimes sti pulated bac kg round informa tion is needed to g et the rig ht 7 re su lt o f t he tho ug ht e xpe ri me nt. Th e a ns we rs we g ive to c a no nic a l qu e sti on s in c e rt a in thought e x periments ar e af fec ted by what is true a bout the actua l world. I will arg ue below that the zombie thought e x periment is one in which the re is re levant informa tion about the ac tual world that may aff ect the a nswer one g ives to a ca nonical question; c on se qu e ntl y , p a rt ic ipa nts c on tin ua lly g ive the wr on g a ns we r t o th e c a no nic a l qu e sti on in

PAGE 186

177 See 8 Chalmers 1996, 94-99. There are other thoug ht experiments that philos ophers use as evide nce tha t 9 phy sicalism is false. See, f or example, J ackson 1998b. Especia lly since ther e ar e such a rang e of r easona ble re actions to the " No" 10 response to the canonic al question. the zombie thought experiment. Althoug h the thoug ht experiment is properly perf ormed it cannot be suc cessf ully perf ormed unless the pa rticipant has the r eleva nt empirical information. M ap o f Re spo nse s t o t he Z om bie Thou ght Expe r im e nt Results of thought experiments sometimes tell us imm ediately whether a conc ept applies or f ails to apply to an imag inary situation. That is, once we imag ine a situation and answ er the rele vant ca nonical question, sometimes we c an say something a bout the conce pt in question (e.g ., whether it applies, fails to apply , or it is indeterminate whe ther i t a p p l i e s t o t h e t a r g e t s c e n a r i o ) . H o w e v e r , t h o u g h t e x p e r i m e n t s a r e n o t a l w a ys straig htforwa rd in this way even w hen they seem to be a t first g lance . The a nswer to the canonic al question may not settle issues that ge t at the hea rt of the matter in question, and, fur thermore , differ ent par ticipants may have dif fer ent intuitions in answer ing the canonic al question. Many anti-phy sicalists have take n Chalmers' s zom bie thoug ht experiment to be par t of a c onceptua l arg ument ag ainst phy sicalism (Chalmers 1996). 89 They arg ue that since we c an conc eive of zombi es, they must be possible. I f zombies are po ssi ble , th e n p hy sic a lis m mu st b e fa lse . I f t his is t he re su lt o f t he tho ug ht e xpe ri me nt, t h e p h y s i ca l i s t wi l l b e m o t i v at ed t o l o o k fo r s o m et h i n g t h at h as go n e w ro n g. I do no t w a nt t o a rg ue tha t th e zomb ie tho ug ht e xpe ri me nt h a s g on e ra dic a lly wr on g , b ut, ra the r, tha t th ing s a re no t a s st ra ig htf or wa rd a s th e y se e m to be , a nd , a s a 10 result, par ticipants tend to make a cer tain kind of mistake. I nstead of c oncluding that

PAGE 187

178 zombi es ar e possible from the f act that they are conce ivable, some philosophers ha ve denied that c onceiva bility entails possibili ty and ar g ue for phy sicalism in part by g iving an explanation of why zombi es see m to be possible but are not. B ut, then, one mig ht a rg ue , it loo ks a s if c on c e ptu a l a na ly sis ha s le d u s a str a y , s o w e sh ou ld q ue sti on the ro le it p la y s in do ing ph ilo so ph y (o f m ind ). Of te n th is k ind of re sp on se c a n b e tie d to Quinean w orries a bout conce ptual analy sis, which I discuss in Chapter 4 (thoug h not as re sp on se s to the zomb ie tho ug ht e xpe ri me nt pe r se ). Or, one might ar g ue that conc eptual analy sis, and in particular , thought e x periments, play a role in doing philosophy , but sometimes we g et the wr ong result bec ause of some problem with our pe rfor mance of them. As I have indica ted on the map ( Fig ure 4) , there are three scena rio intuiti ons one ma y ha ve in r e sp on se to t he c a no nic a l qu e sti on a sk e d in the zomb ie tho ug ht e xpe ri me nt: (a) No, it could fail to be c onscious, (b) Ye s, it must be consc ious, and (c) We are not in a po sit ion to s a y . E a c h o f t he se re sp on se s se e ms t o b e a pr op e r s c e na ri o in tui tio n, tha t is , an intuition that a suitable candida te for pa rticipation in a thoug ht experiment may have. I n th e re st o f t his c ha pte r, I wi ll f ir st s pe ll o ut s ome imp lic a tio ns of the se sc e na ri o int uit ion s a nd the n g o in to g re a te r d e ta il a bo ut t he m. Mo st p a rt ic ipa nts wi ll l ike ly re sp on d to the c a no nic a l qu e sti on in t he zomb ie thought e x periment by answe ring that the being could fa il to be conscious. That is, we can c onceive of a be ing tha t is functionally identical to us, etc., w hich fa ils to have

PAGE 188

179 I should point out that one could ha ve the intuition that the being cannot be 11 conscious, thoug h I do not think any one does, since they would then have to think they themselves ar e not consc ious. The anti-phy sicalist only needs the intuiti on that the being c ou ld fa il t o b e c on sc iou s f or the tho ug ht e xpe ri me nt t o b e us e d a s e vid e nc e a g a ins t ph y sic a lis m. There are many anti-phy sicalist views, but to talk about eac h of them 12 ind ivi du a lly wo uld be a dis se rt a tio n b y its e lf . T o r e a d a bo ut s ome a nti -p hy sic a lis t vi e ws , see Chalmers 1996. conscious experienc e. I t is fairly straig htforwa rd to see how answe ring "N o, the being 11 could fa il to be conscious" to the canonic al question ca n allow us to infer tha t phy sicalism is false. I f, g iven the tar g et sce nario, it looks as if the being need not be 12 conscious, then one can c onclude that zombies are conce ivable. I f zombies are conce ivable, then, one may arg ue, zombies are possible. I f zombies are possible, then phy sicalism is false. F or if ther e is a c omplete phy sical desc ription that leaves phenomena l experience out, then a complete phy sical duplicate of our wor ld may not be a duplicate simpliciter . I t is less straightfor war d to see how a nswering in this way can still allow us to defe nd phy sicalism. The phy sicalist must t ake se riously the intuition t hat the be ing de sc ri be d in the ta rg e t sc e na ri o n e e d n ot b e c on sc iou s d e sp ite be ing ph y sic a lly ide nti c a l to us , s inc e on e ma y e a sil y us e it t o a rg ue a g a ins t ph y sic a lis m. I n f a c t, explaining the appa rently neg ative re sult of the zombi e thoug ht experiment is one of the main obstacles that phy sicalists must overcome. There are a number of wa y s that one ca n explain what has g one wr ong with the zomb ie tho ug ht e xpe ri me nt. Som e ph ilo so ph e rs a rg ue tha t g e ne ra lly the re a re pr ob le ms with doing thoug ht experiments by arg uing a g ainst the notion of conce ptual truth or the analy tic-sy nthetic distinction, or by arg uing tha t once we see the distinction between conce ptual and metaphy sical possibilit y , we will see that thought e x periments just are not

PAGE 189

180 See, for example, 13 Smart 1959. Of c ourse the inc onceiva bility of zombies alone does not g et us phy sicalism, but 14 it d oe s a vo id o ne ma jor ob je c tio n to ph y sic a lis m. re le va nt t o me ta ph y sic a l is su e s. Ot he rs a rg ue tha t th e zomb ie tho ug ht e xpe ri me nt m us t be tre ated diff ere ntly than others f or var ious reasons or that there are few or no interesting a priori truths that we can unc over by doing thoug ht experiments. St ill others arg ue that ther e is a proble m with the zom bie thoug ht experiment beca use we do not succe ssfully perf orm it. The sec ond and third possible response s to the canonic al question, " Yes, the being must be conscious" and " I am not in a position to say ," do not lea d us in as many dir e c tio ns a s th e " no " re sp on se . O ne wh o h a s th e sc e na ri o in tui tio n th a t th e be ing mus t be c on sc iou s w ill lik e ly c on c lud e tha t zom bie s, the re fo re , a re inc on c e iva ble , a nd , th us , impossibl e. I f this is the corre ct intuition, t hen the phy sical wa y the world is entails the phenomena l way the world is, and phy sicalism is true. Even be tter, we g et to the truth of ph y sic a lis m a pr ior i. Ho we ve r, thi s in tui tio n is c on tr a ry to m os t in div idu a ls' sc e na ri o 13 int uit ion s a bo ut t he zomb ie tho ug ht e xpe ri me nt. So, on e wh o h a s th is i ntu iti on sh ou ld d e f e n d i t i n s o m e w a y. 14 One who ha s the scena rio intuiti on that she is not in a position t o answe r the canonic al question may think that given w hat she knows, she is not sure whethe r the descr iption is consistent with what the ca nonical question is asking . That is, she may arg ue that for all we know, w e just are not able to g ive an a nswer to the c anonica l question. Depe nding on how one re ads the third re sponse, it may seem to collapse into a varia tion of the " no" a nswer, sinc e if one answe rs, " I am not in a position to say ," she can e a sil y a dd , " so , I g ue ss i t' s po ssi ble that the being not be consc ious" (tha t is, given tha t

PAGE 190

181 I t is as a re sult of this debate, I think, that philos ophers ha ve eve n questioned the 15 p r i n c i p l e t h a t c o n c e i v a b i l i t y e n t a i l s p o s s i b i l i t y. we do not know wha t to say , it does not look as if the being must be conscious). Howeve r, as I stated, it looks as if "I am not in a position to say " c an stand a lone as a n a ns we r t o th e c a no nic a l qu e sti on if we thi nk a bo ut i t a s b e ing a re sp on se a bo ut w ha t is epistemically possible. P r obl e m s wit h Co nc e pt ual Ana lys is There are a number of implications of the " no" a nswer to the c anonica l question in addition to the standard anti-phy sicalist one. A phy sicalist will li kely arg ue that a lth ou g h zo mbi e s a re c on c e iva ble , th e y a re no t po ssi ble . Si nc e a nti -p hy sic a lis m is tr ue if zombi es ar e possible, we mig ht arg ue that they are not possible, and phy sicalism is true. I find this kind of answer the most interesting , since issues about the r ole of c onceptua l analy sis arise her e. Doing thought e x periments is how we do conce ptual analy sis and answe r metaphy sical questions. Historically , we ha ve take n the re sults of thought e xpe ri me nts , w hic h h e lp u s d e te rm ine wh a t is c on c e iva ble , a s e vid e nc e fo r w ha t is possible, and, thus, as evidenc e for or ag ainst philosophical theories. Tha t is, if we deter mine that something is conc eivable a fter doing a thoug ht experiment, we there by conclude that it is poss ible. I f zombies are c onceiva ble, but not possible, then one 15 needs to e x plain why . I n doing so, one has to say something a bout conce ptual analy sis—either we have be en mistaken a bout the conclusions we ha ve dra wn from thought e x periments, and, so, the ir results should be set aside , or they provide e vidence of some sort that is rele vant to metaphy sical issues, but more ne eds to be said. There are at least two re asons why one mig ht deny that the conc eivability of zombi es does not enta il the possibil ity of them: (a) in ge nera l, the results of thoug ht

PAGE 191

182 I f we want to use possible worlds talk, one who holds this kind of view thinks 16 tha t th e se t of me ta ph y sic a lly po ssi ble wo rl ds is a su bs e t of c on c e ptu a lly po ssi ble wo rl ds . experiments are not releva nt to metaphy sical issues, and ( b) the re sults of thought experiments are rele vant to metaphy sical issues, but we g et the wr ong result in the zomb ie tho ug ht e xpe ri me nt. I n th is s e c tio n, I wi ll d isc us s p os iti on s th a t c ome a bo ut a s a re su lt o f t hin kin g (a ). I n th e ne xt se c tio n, I wi ll d isc us s p os iti on s th a t c ome a bo ut a s a result of thinking (b). I f we have be en mistaken a bout the conclusions we ha ve dra wn from thoug ht e xpe ri me nts , a nd , c on se qu e ntl y , th e re su lts of tho ug ht e xpe ri me nts a re no t r e le va nt t o metaphy sical issues, then ther e ar e ser ious problems with the way we ha ve bee n doing philosophy . There are at least four kinds of positions t hat ena ble philosophers to arg ue this way . I will discuss each br iefly in turn. First, someone mig ht arg ue that ther e ar e tw o k ind s o f p os sib ili ty , c on c e ptu a l po ssi bil ity a nd me ta ph y sic a l po ssi bil ity . F or tho se wh o h old thi s k ind of vie w a bo ut p os sib ili ty , s ome thi ng ' s b e ing c on c e ptu a lly po ssi ble is not sufficient for its being me taphy sically possible. When participating in a thoug ht 16 experiment, they interpre t the canonic al question as one a bout conce ptual possibili ty . Th e kin d o f p os sib ili ty tha t w e a re int e re ste d in in t he zomb ie tho ug ht e xpe ri me nt, howeve r, is metaphy sical possibilit y . So, the results of the zombie thought experiment need to be set aside unless we arg ue that bec ause zombies are conce ptually possible, they are metaphy sically possible. Se c on d, on e mig ht a rg ue tha t a lth ou g h th ou g ht e xpe ri me nts c a n il lum ina te c on c e ptu a l tr uth s, the re a re no c on c e ptu a l tr uth s. Ra the r, e ve ry tr uth is s y nth e tic . So me ph ilo so ph e rs ha ve un de rs too d Q uin e (1 98 0) a s a rg uin g thi s w a y . I re fe r t o th e se individuals as quasi-Quinea ns, since I think Quine arg ues instead that the distinction

PAGE 192

183 betwee n the ana ly tic and sy nthetic is illusory . So, another wa y one mig ht expl ain why t h e r es u l t s o f t h o u gh t ex p er i m en t s ar e n o t re l ev an t t o m et ap h y s i ca l i s s u es i s b y t ak i n g a Quinean line of arg ument: any sense we try to make of the status of those truths, or the appea ranc e of a distinction between them, is g rounded in some c onfusion. Along other qu a siQu ine a n li ne s, so me on e mig ht a rg ue tha t th e pu rp os e of tho ug ht e xpe ri me nts is t o a c qu ir e kn ow le dg e of c on c e ptu a l tr uth s, bu t th e tr ou ble is t ha t c on c e ptu a l tr uth s a re a ll a posteriori, so thoug ht experiments cannot y ield information about them. One who holds thi s p os iti on mig ht a rg ue tha t a ll t he re is t o me a nin g is r e fe re nc e , s o th e on ly wa y to deter mine truth in virtue of " meaning " is to investig ate e mpirically what re fer s to what. This individual mi g ht arg ue that the distinction betwee n the ana ly tic and sy nthetic ca n survive, but conc eptual truths ar e a poste riori; howeve r, it is a bit unclear what the r eal distinction would be. P roblem s w ith the Zom bie Th ought Exp erim ent and/or the Concept of Cons c iou sne ss I f t he re su lts of tho ug ht e xpe ri me nts pr ov ide e vid e nc e of so me so rt tha t is rele vant to metaphy sical issues, but the zombie thought experiment does not provide tha t evidenc e, then we must be clea r about the c onditions under which the r esults of thoug ht experiments are not releva nt to metaphy sical issues. That is, the phy sicalist who (a) answe rs " no" to the c anonica l question, but (b) does not think that zom bies are po ssi ble , y e t ( c ) t hin ks tha t in g e ne ra l th e re su lts of tho ug ht e xpe ri me nts a re re le va nt t o me ta ph y sic a l is su e s, ne e ds to g ive a n a c c ou nt o f t he c on dit ion s u nd e r w hic h th e re su lts a re re le va nt. Al tho ug h it tur ns ou t th a t mo st o f t he tim e wh e n w e pr op e rl y pe rf or m a thought e x periment we g et the rig ht result, sometimes we do not. So, the results of thought e x periments ar e re levant to metaphy sical issues if and only if the re sults are

PAGE 193

184 arr ived at via a succe ssful perf ormanc e of the thought e x periment. One might ar g ue that the zombie thought experiment is one that must meet a fur ther c ondition for it to be su c c e ssf ul. Ei the r t he tho ug ht e xpe ri me nt c a nn ot b e su c c e ssf ull y pe rf or me d b e c a us e there is something exceptional about it, and the r esults it y ields are not trustworthy , or the participa nt in the thought experiment must have f urther e mpirical informa tion about the actua l world in order to ge t reliable r esults beca use the c oncept of c on sc iou sn e ss fu nc tio ns lik e a na tur a l ki nd c on c e pt, or so me thi ng e lse . I n th e fo llo wi ng se c tio n, I wi ll discuss the option acc ording to which the zombie thought experiment ca nnot be su c c e ssf ull y pe rf or me d b e c a us e the re is s ome thi ng or oth e r e xce pti on a l a bo ut i t. Exceptionalist Analys es Some phy sicalists think t hat the re ason why we g et anti-phy sicalist results from the zombie thought experiment is that the conce pt of c on sc iou sn e ss (and othe r phenomena l conce pts) is ex ceptional ( i.e., it is differe nt from other c oncepts). Since c on sc iou sn e ss has ce rtain unique fe ature s, they arg ue, thoug ht experiments involvi ng the conce pt are likely to misl ead us. The re a re a number of w ay s philosophers have a rg ued that consciousness is exceptional: (a) phenomena l conce pts have a specia l nature, ( b) we a re c og nit ive ly c los e d to the re le va nt f a c ts o r c on c e pts in t his c a se , o r ( c ) t he me c ha nis m of conc eiving in this ki nd of ca se is unreliable . Since the conc ept of c on sc iou sn e ss is specia l in some way , some philosophers ar g ue, we cannot trust the r esults we g et from imag ining the zombie thoug ht experiment and others involving the conc ept of c on sc iou sn e ss . They arg ue that althoug h other thoug ht experiments dealing with other c on c e pts a re un pr ob le ma tic , th e zomb ie tho ug ht e xpe ri me nt m us t be tr e a te d d if fe re ntl y (i.e., we should not use the results we g et from a s evidenc e for or ag ainst a philosophical theory ).

PAGE 194

185 L oar ( 1990) ar g ues that wha t is unique about phenomena l conce pts is that they are rec og nitional concepts, a nd there is no property distinct from the phenomena l proper ty itself "c onnoted" as a c ontinge nt mode of pre sentation. Consequently , they direc tly pick out the phy sical prope rties to which they ref er. Tha t is, there is no dis tin c tio n b e tw e e n th e ph y sic a l pr op e rt y the ph e no me na l c on c e pt r e fe rs to a nd ho w i t is prese nted to the subject in experienc e. Rec og nitional concepts " have the form ' x is one of that ki nd '; t hey are t y pe-d emo ns tr ati ves " (L oa r 1 99 0, 60 0) . I n o the r w or ds , re c og nit ion a l c on c e pts a re c on c e pts fo r w hic h th e re a re no a sso c ia te d d e sc ri pti on s a t a ll; they are just modes of prese ntation. These ty pe demonstra tives are g rounded in dispositions to classify , by way of p er ce p t u al d i s cr i m i n at i o n s , ce rt ai n o b j ec t s , ev en t s , s i t u at i o n s . S u p p o s e y o u go int o th e Ca lif or nia de se rt a nd sp ot a su c c ule nt n e ve r s e e n b e fo re . Y ou be c ome adept a t rec og nizi ng instances, a nd g ain a r ecog nitional command of their kind, without a name f or it; y ou are disposed to identify positive and neg ative instance s a nd the re by pic k o ut a kin d. Th e se dis po sit ion s a re ty pic a lly lin ke d w ith c a pa c iti e s to fo rm ima g e s, wh os e c on c e ptu a l r ole se e ms t o b e to f oc us tho ug hts about an identifia ble kind in the absenc e of c urre ntly perc eived instanc es. (L oar 1990, 600) Con se qu e ntl y , th e re is n o a pr ior i c omp on e nt t o th e me a nin g of the te rm s th a t e xpre ss rec og nitional concepts. F or example, the conc ept of pa in direc tly picks out the phenomena l experience of being in pain, so there is nothing we c an say a priori a bout pain. We become adept a t rec og nizi ng when we are in pain, but there is nothing we c an say a priori a bout being in pain. Since there is no a priori compone nt to the meaning s of terms that express phenomena l conce pts, doing thoug ht experiments on them is fruitless. Mc Gi nn (1 99 8) a rg ue s th a t th e re su lts of the zomb ie tho ug ht e xpe ri me nt s ho uld no t be us e d a s e vid e nc e fo r o r a g a ins t ph y sic a lis m be c a us e we a re c og nit ive ly c los e d to the re le va nt f a c ts o r c on c e pts in t his c a se :

PAGE 195

186 we a re c ut off by our ver y cog nitive constituti on from ac hieving a conc eption of that natura l property of the bra in (or of c onsciousness) that a ccounts for the psy chophy sical link. This is a kind of causa l nexus that we a re pr eclude d from ever understanding , g iven the wa y we ha ve to form our conce pts and deve lop our theories. (McGinn 1998, 529) McGinn explains what he mea ns by being "c ut off" from under standing consciousness by invoking the notion of c og nitive closure: A ty pe of mind M i s c o g n i t i v e l y c l o s e d w i t h r e s p e c t t o a p r o p e r t y P ( o r t h e o r y T ) if a nd on ly if the c on c e ptfo rm ing pr oc e du re s a t M ' s disposal cannot extend to a g rasp of P (or a n understanding of T ). (McGinn 1998, 529) As a re sult of being cog nitively closed to the prope rty of consc iousness, he ar g ues, we cannot unde rstand how c onsciousness is relate d to the phy sical. Ac cording to McGinn, the re a re thr e e po ssi ble wa y s th a t th e min d a nd bo dy a re re la te d: ( a ) t he re is s ome br ute fac t about how consc iousness comes about, ( b) there is some miraculous occ urre nce tha t happens w hen ther e is consciousne ss, or (c) there is some natural a ccount of c on sc iou sn e ss. He c la ims tha t th e fi rs t tw o a re imp la us ibl e a nd c on c lud e s th a t th e re mus t be some na tural ac count of c onsciousness by the proc ess of e limination. The struc ture of Mc Ginn' s arg ument for the c laim that we ar e cog nitively closed wi th r e sp e c t to the pr op e rt y of c on sc iou sn e ss i s a s f oll ow s: (1) We can unde rstand P either by introspection or by study ing the brain. (2 ) I ntr os pe c tio n d oe s n ot r e ve a l P . (3 ) Stu dy ing the br a in d oe s n ot r e ve a l P . (4) There fore , we c ould never arr ive at a g rasp of P . I t is not i mportant for my purposes to g o into the details of how he justifies (3) and (4) . So, fo r M c Gi nn (1 98 4) , th e re is n o mi nd -b od y pr ob le m pe r se . Co ns c iou sn e ss i s ph y sic a l, b ut t he re is n o w a y fo r u s to kn ow ho w i t is .

PAGE 196

187 Th e y us e the e xpre ssi on ' re c og nit ion a l c on c e pt' , b ut i t is no t c le a r t ha t th e y us e it 17 in t he sa me wa y L oa r ( 19 90 ) d oe s. Hill and McL aug hlin (1999) ar g ue both that phenomena l conce pts are re c og nit ion a l c on c e pts a nd tha t th e re a re tw o f or ms o f i ma g ina tio n th a t w e ma y us e 17 wh e n p a rt ic ipa tin g in a tho ug ht e xpe ri me nt, pe rc e ptu a l im a g ina tio n o r s y mpa the tic imag ination (or both). The y borrow the distinction from Nag el (1974): We m a y ima g ine so me thi ng by re pr e se nti ng it t o o ur se lve s e ith e r p e rc e ptu a lly [or] sy mpathetically . . .To imag ine something perc eptually , we put ourse lves in a conscious state r esembling the state we would be in if we perc eived it. To ima g ine so me thi ng sy mpa the tic a lly , w e pu t ou rs e lve s in a c on sc iou s st a te re se mbl ing the thi ng its e lf . ( Th is m e tho d c a n b e us e d to ima g ine me nta l e ve nts a nd sta te s— ou r o wn or a no the r' s. ) Wh e n w e tr y to i ma g ine a me nta l st a te occur ring without its associate d brain state, w e first sy mpathetically imag ine the oc c ur re nc e of the me nta l st a te : th a t is , w e pu t ou rs e lve s in a sta te tha t r e se mbl e s it mentally . At the same time, we attempt to perc eptually imag ine the nonoc c ur re nc e of the ph y sic a l st a te . Wh e re the ima g ina tio n o f p hy sic a l f e a tur e s is perc eptual and the imag ination of mental fe ature s is sy mpathetic, it appea rs that we c an imag ine any experience occur ring without its associate d brain state, a nd vic e ve rs a . T he re la tio n b e tw e e n th e m w ill a pp e a r c on tin g e nt e ve n if it i s nece ssary , beca use of the inde pendenc e of the disparate ty pes of imag ination. (Na g el 1974, 527 n. 11) When participating in the zombi e thoug ht experiment, we ar e supposed to conc eive of a being that is in the same bra in states we a re in, but which is not consc ious. This requires e xer c isi ng bo th p e rc e ptu a l im a g ina tio n a nd sy mpa the tic ima g ina tio n. Ev e n if we a ssu me that the two kinds of imag ination are reliable when we exercise them indepe ndently , they may not be when w e exerc ise them tog ether : Con c e ivi ng of a sit ua tio n in wh ic h o ne is i n a se ns or y sta te wi tho ut b e ing in i ts nomologica lly corr elated br ain state will involve the joint ex erc ise of a se nsory and a phy sical conc ept. . . . Sensory states ar e selfprese nting state s: we e xp e r ie n c e th e m, b u t w e d o n o t h a v e s e n s o r y e xp e r ie n c e s o f th e m. We experience them simply in virtue of being in them. Sensory conce pts are re c og nit ion a l c on c e pts : de plo y ing su c h c on c e pts , w e c a n in tr os pe c tiv e ly re c og nize wh e n w e a re in s e ns or y sta te s si mpl y by fo c us ing ou r a tte nti on dir e c tly on them. Matters a re of course quite differ ent in the ca se of pe rce ptual and theore tical conc epts. An ag ent' s acc ess to the phenomena that he or she perc eives is alway s indirect: it alway s occur s via an e x perie nce of the per ceive d phenomena

PAGE 197

188 that is not identical with the perc eived phe nomenal, but ra ther c aused by it. The acts of c onceiva bility that are responsible for Cartesian intuitions involve the joint exercise of sensory and phy sical conc epts. Modal intuitions resulting from the joint ex erc ise of phy sical conc epts alone or sensory conce pts alone could be reliable , while Cartesia n intuiti ons are sy stematically misleading . Given the wa y in w hic h th e fu nd a me nta l e pis te mic c on str a int s o n th e us e of se ns or y c on c e pts differ from those on the use of phy sical conc epts, there is no a p rio ri re a so n to trust such intuiti ons. Given these dif fer ence s betwee n sensory conce pts and ph y sic a l c on c e pts , a se ns or y sta te a nd its no mol og ic a lly c or re la te d b ra in s ta te would seem conting ently rela ted, eve n if there wer e nec essar ily one. (Hill and McL aug hlin 1999, 448-449) Be cause we a re e x erc ising two diff ere nt kinds of imagina tion in the z ombie thought experiment, it may seem a s if zombi es ar e possible when, in f act, they are not. Hill and Mc L a ug hli n ( 19 99 ) w ou ld s a y tha t th e re a so n w e te nd to h a ve Ca rt e sia n ( tha t is , d ua lis t) int uit ion s w he n w e ima g ine the zomb ie ta rg e t sc e na ri o is tha t w e a re us ing bo th o f t he se kinds of imag ination at the same time, but we have no r eason to trust them tog ether . So, we should not rely on our intuitions i n the zombie thought experiment bec ause two kinds of imag ination are at play when we imag ine the tar g et sce nario a nd answe r the c anonica l qu e sti on . T he zomb ie tho ug ht e xpe ri me nt, a s a re oth e rs inv olv ing ph e no me na l c on c e pts , is e xce pti on a l. Ascr iptive and Non-Ascr iptive Approaches Th e re a re so me ph ilo so ph e rs wh o th ink tha t c on sc iou sn e ss is a natura l kind c on c e pt, a nd tha t th a t se rv e s a s p a rt of the re a so n w hy we a re in e rr or a bo ut t he zomb ie thought e x periment. How ever , there are at least thre e wa y s that one may arg ue that issues about a poster iori nece ssity are rele vant her e. One may arg ue, as I do, that c on sc iou sn e ss is a na tur a l ki nd c on c e pt a nd the sta nd a rd a c c ou nt o f a po ste ri or i ne c e ssi ty that I discussed ea rlier is cor rec t. I will discuss thi s analy sis of the zombi e thoug ht experiment below. Alterna tively , one may arg ue that phenome nal conc epts function like na tur a l ki nd c on c e pts , b ut t ha t th e sta nd a rd a c c ou nt o f a po ste ri or i ne c e ssi ty is i nc or re c t,

PAGE 198

189 I bo rr ow the te rm s ' a sc ri pti ve ' a nd ' no na sc ri pti ve fr om 18 L evine 2001. y et there still is a problem with the zombie thought experiment. Ther e ar e two re asons why one may arg ue that as a result the zombie thought experiment is unreliable : (a) thought e x periments ca n help us find a pr iori conce ptual truths, but there a re ve ry few of the m, t oo fe w t o ma ke the sta nd a rd e xpla na tio n p la us ibl e , o r ( b) tho ug ht e xpe ri me nts can only help us find a pr iori conce ptual truths, but there a re none to be found, whic h makes the standa rd explanation implausible. I call (a ) and ( b) the a scriptive and n o n a s c r i p t i v e a p p r o a c h e s , r e s p e c t i v e l y. 18 On e ma y thi nk tha t c on sc iou sn e ss is a natura l kind concept, without thinking that a dd ing fu rt he r e mpi ri c a l in fo rm a tio n a bo ut t he a c tua l w or ld w ill g e t th e pa rt ic ipa nt t o reinter pret wha t she is conce iving. I n fac t, adding stipulated backg round informa tion to a tho ug ht e xpe ri me nt i s n ot r e le va nt ( tho ug h it c ou ld be—bac kg round informa tion in fact jus t ne ve r i s r e le va nt b e c a us e the re is n e ve r e no ug h c on c e ptu a l kn ow le dg e to m a ke it rele vant) to an a scriptivist because thought e x periments ar e not re levant to deter mining whether a conc ept applies, fa ils to apply , or it is indeterminate whe ther it applies. The ascr iptivis t arg ues that we think we have a priori knowle dg e of f acts a bout how the ref ere nts of conce pts are de termined. How ever , althoug h there are a priori truths couche d in terms of our c oncepts, ther e ar e not many of them, and, w ith respec t to natural kin d c on c e pts , th e tr uth s a re no t r ic h e no ug h to g ive us a ns we rs to t ho ug ht e xpe ri me nts involving them. With respect to the zombie thought e x periment, the inter esting c on c e ptu a l tr uth s a bo ut c on sc iou sn e ss are not rich enoug h to link them to phy sical conce pts. So, t he zombie thought experiment see ms to show that phy sicalism is false. B ut, the a sc ri pti vis t a rg ue s, we sh ou ld b e wa ry of thi s r e su lt.

PAGE 199

190 The non-a scriptivist, li ke the a scriptivist, does not think t hat adding empirica l information will help us succe ssfully perf orm the zombie thought experiment. Unlike the ascr iptivis t, however , the non-a scriptivist never thinks that such bac kg round informa tion is at all, in principle, eve n possibly rele vant (ther e is no conc eptual knowledg e of the rele vant sort to exploi t that backg round informa tion). Every thing we know about our conce pts is a posteriori. As a r esult, then, the zombie thought experiment does not show us that phy sicalism is false. To show that phy sicalism is false we would need e mpirical evidenc e. The purpose of the last fe w sec tions was to lay out some responses tha t phy sicalists have g iven to the zombie thought experiment, none of w hich I find sa tis fy ing . E a c h o f t he m ta ke fo r g ra nte d th e a nti -p hy sic a lis t r e su lts of the zomb ie thought e x periment or invoke some kind of skepticism about conc eptual ana ly sis. I n the next s ection, I provide phy sicalists with a response to the zombie thoug ht experiment that explains why we may have a nti-phy sicalist intuit ions about it that makes use of the appar atus I have introduc ed ea rlier in this dissertation. Ne gle c te d P hys ic ali st Re spo nse to th e Z om bie Thou ght Expe r im e nt I n th is s e c tio n, I a pp ly the a pp a ra tus I ha ve de ve lop e d e a rl ie r t o th e zomb ie tho ug ht e xpe ri me nt. I n c on tr a st t o o the r a pp ro a c he s to e xpla ini ng the a pp a re nt a nti phy sicalist results of the zombie thought experiment, I arg ue that the zombie thought experiment can be criticized using the resour ces I have de veloped. I n particula r, I arg ue tha t on e a pp e a lin g wa y fo r t he ph y sic a lis t to a tta c k th e us ua l r e su lt o f t he e xpe ri me nt i s by arg uing tha t the participa nt is insufficiently sensitive to the way s in which bac kg round information about the a ctual wor ld can in fa ct be r eleva nt to the answer to the canonic al

PAGE 200

191 I int ro du c e thi s k ind of pr ob le m in my dis c us sio n o f P utn a m' s T wi n E a rt h 19 thought e x periment, a nd I will discuss it briefly ag ain her e. Contrary to popular opinion I sug g est in this dis sertation that consc iousness be 20 tr e a te d a s a na tur a l ki nd c on c e pt. Th is, of c ou rs e , is c on tr ov e rs ia l. A lth ou g h I c ite so me important simil arities betwe en the c oncept of water and the c oncept of c on sc iou sn e ss , I do not offer an ar g ument that shows that the conc ept of c on sc iou sn e ss is a natura l kind c on c e pt. question, and, as a result, the thoug ht experiment is unsuccessfully perf ormed. Such 19 insensitivit y is due to a fa ilure to have been pr esente d with a var iety of thoug ht experiments involving the same conce pt, where those experiments differ prec isely on wh a t th e y pr e su me a bo ut t ha t ba c kg ro un d in fo rm a tio n. I of fe r t his to t he ph y sic a lis t a s a way of re jecting the usual re sult without having to become mor e g ener ally skeptical a bo ut t ho ug ht e xpe ri me nts . Th e a pp lic a tio n o f t he a na ly sis pe r se of the zomb ie tho ug ht e xpe ri me nt j us t shows what ar e the possible sourc es of e rror in the thought e x periment. B ut by examination of a para llel case about which ther e is broad a g ree ment, a var iation of Putnam's Twin Ea rth thoug ht experiment, I will show that there is one pa rticular kind of fa ilu re to w hic h th ou g ht e xpe ri me nts of jus t th e zomb ie so rt a re pr on e . Ad mit te dly , th is 20 fa lls sh y of sh ow ing tha t pe rf or ma nc e s o f t he zomb ie tho ug ht e xpe ri me nt do ty pic a lly fa il in this way . But that is not a proble m for me he re, since the more mode st aim of the curr ent sec tion will be accomplished, na mely , the aim of showing that the ana ly sis has reve aled a serious ave nue of phy sicalist response to Chalmer s' s zom bie thoug ht experiment (Chalmers 1996). As I sta te d a bo ve , s ome ph ilo so ph e rs thi nk tha t th e re a so n w e g e t a nti -p hy sic a lis t results from the zombie thoug ht experiment is that the concept of c on sc iou sn e ss , or phenomena l conce pts in ge nera l, function like natura l kind concepts. I f phenome nal

PAGE 201

192 I should point out here that althoug h the standar d acc ount of a poster iori 21 nece ssity is controver sial, there a re both phy sicalists and anti-phy sicalists who acc ept it. conce pts function like natura l kind terms, as I think we have rea son to believe, then the w a y t h e a c t u a l w o r l d i s a t l e a s t p a r t l y d e t e r m i n e s w h a t f a l l s u n d e r t h e m . C o n s e q u e n t l y, the re sults of the zom bie thoug ht experiment as usually perf ormed a re unr eliable. Doing solely a priori a naly sis of phenomena l conce pts miss es out on an important fe ature of t h e m . O n e m a y a r g u e , a s I d o , t h a t t h e s t a n d a r d e x p l a n a t i o n o f a p o s t e r i o r i n e c e s s i t y I dis c us s in Cha pte r 2 is c or re c t, a nd tha t th e re a so n w hy we a re mis le d in the zomb ie 21 thought e x periment is that when w e ar e imag ining our twin zom bie, we do not ha ve the rele vant empirica l information that may be re quired to g ive an a dequate answe r to the canonic al question. Those who think that the standa rd explanation of a poster iori nece ssity is corre ct may arg ue that the r eason w hy the zombie thought experiment g oes wrong is that the pa rt ic ipa nt d oe s n ot h a ve the re le va nt e mpi ri c a l in fo rm a tio n th a t w ou ld e na ble he r t o conce ive of the r ight thing when she imag ines the targ et sce nario. As a result, it looks as if zombi es ar e conc eivable. H oweve r, if the pa rticipant had the empirica l information tha t th e pr op e rt y tha t pl a y s th e pa in r ole , f or e xamp le , is the pr op e rt y of be ing in a c e rt a in neurolog ical state, the n she would re conce ive the zombie scena rio and c hang e her answe r to th e cano ni cal qu est io n t o "Yes , t he b ein g mus t b e con sci ou s. " The phy sicalist who does not want to re ject the zombie thought e x periment ou tr ig ht h a s to a rg ue tha t so me thi ng ru ns int e rf e re nc e in o ur pe rf or ma nc e of it. Th os e who fa vor the standa rd ac count will arg ue that wha t runs interfe renc e is our lac k of empirica l knowledg e—in a se nse, we just do not know what phy sical prope rty the conce pt of be ing in p ain (or in any other phe nomenal state) picks out. I will draw a n

PAGE 202

193 a na log y be tw e e n th e zomb ie tho ug ht e xpe ri me nt a nd a va ri a tio n o f P utn a m' s T wi n E a rt h thought e x periment, whic h I discussed in Chapter 3. P arallel Cases and P arallel Error s Every one will ag ree that improperly perf ormed thoug ht experiments cannot be trusted, but once w e have a prope rly perf ormed thoug ht experiment, they are much more likely to be convincing , and our intuitions are likely to provide us with g ood evidenc e for or a g a ins t a pa rt ic ula r t he or y . E a rl ie r, I e xpla ine d th e kin ds of mis ta ke s w e ma y ma ke in pe rf or min g tho ug ht e xpe ri me nts , a nd ve ry sp e c ia l ki nd of mis ta ke we ma y ma ke in pe rf or min g a va ri a tio n o f P utn a m' s T wi n E a rt h th ou g ht e xpe ri me nt. I a rg ue in t his section that whe n we pe rfor m the zombi e thoug ht experiment, we make the same kind of mistake. I f the kind of e rror I arg ue we may be making in perfor ming the zombie thoug ht experiment can be avoided, then the mere fac t that they can be made doe s not give us rea son to cea se using thought e x periments altog ether . I f these e rror s cannot be avoided, then we should be awa re of them and avoid using those thoug ht experiments as a wa y of rea ching metaphy sical conc lusions. I f we make the e rror s ever y time we try to do a thought e x periment, then w e have rea son to think t hat we should not use them. I arg ue tha t th e kin d o f e rr or we ma ke in t he zomb ie tho ug ht e xpe ri me nt i s o ne tha t is a t le a st po te nti a lly a vo ida ble , a nd tha t e ve n if it i s n ot, it i s n ot o ne tha t is sy mpt oma tic of a ll thought e x periments. So, one implication of my analy sis of the zombi e thoug ht experiment is that skepticism about the utili ty of thoug ht experiments can be c ombated. As I say in Chapter 3, the mistake I think we may be making in both a varia tion of Putnam's Twin Ea rth thoug ht experiment and the zombie thought experiment is a much mor e su btl e on e tha n th e oth e rs I dis c us se d. Th e e rr or is o ne tha t a llo ws the zomb ie

PAGE 203

194 See 22 Putnam 1975; Kripke 1980. thought e x periment to be pr operly perf ormed, but not succe ssfully perf ormed be cause the participa nt fails to be sensitive to the re levanc e of ba ckg round informa tion to the application of the rele vant conc ept. One ma y wonder how such e rror s are at all possible, g iven that prope r per formanc e re quires that the pa rticipant fully possesses the re levant c on c e pt. L e t me e la bo ra te on jus t w ha t f ull c on c e pt p os se ssi on e nta ils , a nd the n w e wi ll s ee h o w t h e k n o wl ed ge gr an t ed b y s u ch co n ce p t p o s s es s i o n s t i l l al l o ws er ro rs b y t ak i n g a c los e loo k a t a ve rs ion of Put na m' s T wi n E a rt h th ou g ht e xpe ri me nt. I explain full concept possession in terms of spec ific abilities I think a participa nt in a thoug ht experiment should have. I proposed a nd motivated the following conditions in Ch apt er 3 i n t he s ect io n en ti tl ed, "Su it abl e C and id ate s an d C on cept P os ses si on ": A f ull y po sse sse s c on c e pt C iff (a ) A is able to rec og nize a priori as true s u ff i ci en t l y s i m p l e c o n ce p t u al l y t ru e p ro p o s i t i o n s i n v o l v i n g C a nd oth e r c on c e pts which A fully possesses; and (b) A can sa y acc urate ly (and w ith just ification) about any rele vantly completely descr ibed targ et sce nario whe ther C a pp lie s, fa ils to apply , or it is objectively indeterminate or epistemica lly indeterminate whether C a pp lie s. As I explained in Chapter 3, epistemic indete rminacy is an issue in those case s wher e g ener ally the applica tion conditions of the conce pts (in hy pothetical ca ses) a re de te rm ine d in pa rt by the wa y the a c tua l w or ld i s. Na tur a l ki nd c on c e pts a re se ns iti ve in this way , as Kripke and Putnam arg ue in their re spective se minal work on the issue. 22 22 For example, since in the ac tual world wa ter is H O, something that is not H O in a hy pothetical ca se is not water . A fa ilure to be sensitive to bac kg round informa tion and the possibilit y of epistemic indeter minacy in thought experiments involving na tural kind conc epts ca n lead to intuit ions that we should not use in drawing metaphy sical conc lusions. This l ack of

PAGE 204

195 sensitivity , as I sug g est in my discussion of a var iation of the Twin Ea rth thoug ht ex p er i m en t an d i n t h e r es t o f t h i s ch ap t er , d o es n o t p re v en t o n e f ro m p o s s es s i n g a c on c e pt; it j us t ma ke s u s f a lli ble in s a y ing wh e the r a c on c e pt a pp lie s, fa ils to a pp ly , o r i t is indeterminate that it does. I should point out that I think the results of thoug ht experiments tend to be re liable bec ause posse ssing a conce pt require s that one have the skill t o arr ive at the kind of know ledg e that thoug ht experiments are suppose d to y ield. I f we do not appre ciate the importance of not having the re levant bac kg round information in thought e x periments involving na tural kind conc epts, then our intuitions ma y no t pr ov ide g oo d e vid e nc e fo r m e ta ph y sic a l c on c lus ion s. I n th e c a se of the zomb ie tho ug ht e xpe ri me nt a s it is u su a lly pe rf or me d, I a rg ue tha t w e sh ou ld n ot d ra w a n a nti phy sicalist conclusions. I n what follows, I will ex plain how someone w ho possesses the conce pt of water c a n n e ve rt he le ss f a il t o s e e the re le va nc e of ba c kg ro un d in fo rm a tio n in a ns we ri ng the c a no nic a l qu e sti on . T he n I wi ll e xpla in h ow the sa me thi ng is g oin g on in the zomb ie tho ug ht e xpe ri me nt. Variation of T wi n Earth The following is the variation of the Twin Ear th thought experiment I introduced ear lier, which, I arg ue, is ana logous to the zombie thoug ht experiment. Call thi s TE1: Suppose that somewher e in the g alaxy , there is a planet we will call ' Twin Ear th'. Twin Ear th is very simil ar to Ea rth: people on Twin Ea rth speak a lang uag e that sounds exactly like Eng lish, and for e ach pe rson on Ear th there is a pe rfe ct phy sical duplicate of him or her on Twin Ear th. What Earthlings c all ' water ' on Ea rt h a nd wh a t T wi n E a rt hli ng s c a ll ' wa te r' on Tw in E a rt h s ha re the sa me superf icial qualities: what Twin Ea rthlings c all ' water ' on Twin Ear th looks and ta ste s ju st a s w ha t E a rt hli ng s c a ll ' wa te r' on Ea rt h, wh a t T wi n E a rt hli ng s c a ll ' wa te r' on Tw in E a rt h q ue nc he s th ir st j us t a s w ha t E a rt hli ng s c a ll ' wa te r' on Ea rt h do e s, wh a t T wi n E a rt hli ng s c a ll ' wa te r' on Tw in E a rt h f ill s la ke s a nd oc e a ns jus t as wha t Earthling s call ' water ' on Earth doe s, ' water ' is what Twin Ear thlings say comes out of the sky when it is raining , and so on.

PAGE 205

196 Ca noni c al q ue st ion : I s there w ater on Twin Ear th? As I explained in Chapter 3, the answ er tha t one g ives to the ca nonical question ma y de pe nd a t le a st i n p a rt on wh a t f ur the r i nf or ma tio n o ne ha s a bo ut t he a c tua l w or ld (that is, the context one is in), and, per haps, informa tion about Twin Earth. To make the a rg ume nt I wa nt t o ma ke wi th r e sp e c t to the zomb ie tho ug ht e xpe ri me nt, we mus t reme mber a few things a bout the analy sis of TE1 I g ave e arlier . I arg ue that the pr oper re sp on se to t he c a no nic a l qu e sti on is " I do n' t kn ow ," or " I t is e pis te mic a lly indeterminate ." I t is only afte r we provide the pa rticipant with furthe r bac kg round inf or ma tio n, na me ly , th a t w ha t th e wa te ry stu ff pic ks ou t on Ea rt h a nd on Tw in E a rt h is 2 H O and XY Z , respe ctively , that we c an eliminate e pistemic indeterminac y and answ er the c a no nic a l qu e sti on a de qu a te ly . A s I ha ve pr e vio us ly a rg ue d, on c e g ive n th is information, a c andidate should say that there is no water on T win Earth. I a lso ta lke d e a rl ie r a bo ut c on dit ion a l kn ow le dg e tha t w e ma y ha ve a bo ut a conce pt, which is conce ptual knowledg e, that one ma y have in or der to possess a natura l kind conce pt. For e x ample, assuming we did not know the ultimate composition of 22 water , we mig ht know that if in the ac tual world, wa ter is H O, then wat e r ref ers to H O. Ev e n if we be lie ve we kn ow the fe a tur e s o f t he a c tua l w or ld, a s w e do wi th r e sp e c t to water , that deter mine which prope rty is picked out by a conc ept, we still have the c on dit ion a l kn ow le dg e a s w e kn ow wh a t to sa y if we a re in e rr or a bo ut t ho se fe a tur e s. Of c ourse, in the c ase of c on sc iou sn e ss , we do not know its ultimate composition if it i s a na tur a l ki nd c on c e pt. So, the on ly kin d o f k no wl e dg e we ha ve a bo ut i ts c omp os iti on is c on dit ion a l kn ow le dg e , w hic h ma y be imp lic it. We d o n ot k no w w ha t is true a bout the actua l world in the ca se of c on sc iou sn e ss , an d t h e c o n d i t i o n al k n o wl ed ge we (m a y ) h a ve a bo ut i t is lik e ly imp lic it. Con se qu e ntl y , a pe rf e c tly su ita ble c a nd ida te

PAGE 206

197 Asking participa nts to im ag ine the zombie targ et sce nario with diffe rent 23 backg round informa tion may also help us bring to light any backg round assumptions she ma y ha ve a bo ut c on sc iou sn e ss . may fail to rec og nize the releva nce of it to the z ombie thought e x periment. I t is up to t he experimenter to provide the par ticipant with stipul ated ba ckg round informa tion. W ithout the stipulated bac kg round informa tion, I arg ue that a pa rticipant should just not know what to say about the zombie thought experiment. I illust rate this by prese nting the zombi e thoug ht experiment with differe nt backg round informa tion. I t seems that when we va ry the bac kg round informa tion, the participant g ives vary ing a nswers to the c a no nic a l qu e sti on . I n th e ne xt se c tio n, I wi ll d isc us s th e zomb ie tho ug ht e xpe ri me nt. 23 Z om bie Thou ght Expe r im e nt No w t ha t w e ha ve re vis ite d my e va lua tio n o f T E1 , w e sh ou ld e xami ne the zomb ie tho ug ht e xpe ri me nt a nd se e wh a t ha s g on e wr on g . Si nc e we do no t kn ow wh a t pr op e rt y the conc ept of c on sc iou sn e ss picks out, we a re in the sa me epistemic position as those on Earth in 1750, bef ore the discovery of the a tomic theory , with respec t to the conce pt of water . I t a lso se e ms t ha t in 17 50 , u nle ss w e we re g ive n s e ve ra l e xpe ri me nts va ry ing bo th backg round informa tion on what is actually the ca se and the char acte r of the w ater y stuff in t he sc e na ri o, tha t pe rs on ma y fa il t o s e e tha t ba c kg ro un d in fo rm a tio n is re le va nt a t a ll, and, henc e, she c an fa il to see the conc ept of water as a na tural kind conc ept. This may ve ry we ll b e wh a t is g oin g on in t he zomb ie tho ug ht e xpe ri me nt. Th e fo llo wi ng is t he se tup of the zomb ie tho ug ht e xpe ri me nt I pr e se nte d e a rl ie r i n this chapter. Ca ll it TE2: Suppose that there is a being that is identical to me in all of its relational and nonre la tio na l ph y sic a l pr op e rt ie s. Th e be ing e xhibi ts a ll o f t he sa me be ha vio rs tha t I exhibi t, is functionally identical to me, and is in the same phy sical environme nt tha t I a m in —it s st oma c h g ro wl s a t me a lti me , it is d re sse d in c lot he s th a t ma tc h, it

PAGE 207

198 rea cts in the same wa y that I rea ct to the same stimuli insofar as a ny nonint e nti on a l, n on -p sy c ho log ic a l de sc ri pti on of it g oe s ( e .g ., it s e e ms t o r e po rt on its internal states, e tc.). Ca noni c al q ue st ion : Mu st t he be ing be c on sc iou s? I f TE2 a s perf ormed now is ana logous to TE1 a s perf ormed in 1750, then the answe r one g ives to the ca nonical question should depend on w hat furthe r informa tion one has a bout the actua l world. I f consc iousness is a context-depende nt conce pt, then the ri g ht a ns we r m ig ht b e tha t it is e pis te mic a lly ind e te rm ina te wh e the r t he be ing is c on sc iou s. Of c ou rs e mos t pa rt ic ipa nts do no t r e a c t th a t w a y . M os t su ita ble pa rt ic ipa nts answe r, " No, the being could fa il to be conscious." That is, it seems that we c an conc eive of a be ing tha t is functionally identical to us, etc., w hich fa ils to have conscious e xpe ri e nc e . B ut w hy do e s it se e m th a t w e c a n c on c e ive of thi s k ind of be ing ? I su g g e st for the sa me re ason those par ticipating in TE1 in 1750 will likely say that there is water on Twin Ear th: they have not be en pre sented with thoug ht experiments with vary ing stipulated backg round informa tion, so they are not sensitive to the releva nce of such b a c k g r o u n d in f o r ma ti o n . V a r y in g b a c k g r o u n d in f o r ma ti o n ma y h e lp a p a r ti c ip a n t r e a li ze tha t sh e wa s ma kin g a n a ssu mpt ion a bo ut t he a c tua l w or ld, or it m a y c a us e he r t o re c on sid e r h e r a ns we r t o th e c a no nic a l qu e sti on . I wi ll s pe ll t his ou t in mor e de ta il below. Of course information about the a ctual wor ld is not alway s releva nt to what we wi ll s a y in r e sp on se to t he c a no nic a l qu e sti on in a tho ug ht e xpe ri me nt, a nd , th us , to succe ssful perf ormanc e of a thought e x periment. F or natura l kind concepts, howe ver, what the a ctual wor ld is li ke is rele vant to their applica tion conditions. Many participa nts in the z ombie thought e x periment fa il to recog nize that it m ay be e pis te mic a lly ind e te rm ina te wh e the r t he zomb ie de sc ri be d in the ta rg e t sc e na ri o is c on sc iou s. I f i t is e pis te mic a lly ind e te rm ina te , th e n s inc e we us ua lly sa y tha t th e zomb ie

PAGE 208

199 may not be consc ious, we ar e unsucc essful in our per formanc e of the zombi e thoug ht experiment. I t is thi s possibil ity that I think is an insufficiently appre ciated one in the literature . Successf ul perfor mance of a thoug ht experiment require s not only the pre sence of such bac kg round informa tion, but a sensitivit y to its relevanc e esta blished by the participa nt's ha ving c onsidered a suitably broad r ang e of thoug ht experiments vary ing on 2 that para meter. I f a pa rticipant is told in TE1, for example, that water is H O (or something e lse), then she will likely chang e her response to "I don' t know," or , in other words, it is epistemically indeterminate , beca use she doe s not know if the wate ry stuff on 2 Twin Ear th is also H O. As I stated above , "I don' t know" is the c orre ct answe r to the canonic al question. Simi larly , if we pr ovide the par ticipant in TE2 with stipul ated backg round informa tion, she may chang e her answe r to the ca nonical question, depending what the stipulated ba ckg round informa tion is. For example, suppose we va ry the bac kg round informa tion given to the pa rticipant and a sk her to imag ine the tar g et scena rio first with the stipulated bac kg round informa tion that the thing in the ac tual wo rl d th a t is the e ff e c t of tis su e da ma g e a nd lik e c a us e s is a pr op e rt y , w hic h is , in fa c t, e pip he no me na l; c a ll t his sti pu la te d b a c kg ro un d in fo rm a tio n PE S. T he n w e a sk he r t o pe rf or m it wi th t he sti pu la te d b a c kg ro un d in fo rm a tio n th a t th e thi ng in t he a c tua l w or ld tha t is the e ff e c t of tis su e da ma g e a nd lik e c a us e s is a pr op e rt y , w hic h is , in fa c t, ne ur olo g ic a l; c a ll t his sti pu la te d b a c kg ro un d in fo rm a tio n PN S. Wh e n th e pa rt ic ipa nt i s g iven PES, she will li kely answe r, " No, the being could fa il to be conscious." When the participa nt is given PNS, howeve r, it seems that the pa rticipant will answer , "Y es, the be ing mus t be c on sc iou s. " Sin c e the pa rt ic ipa nt i s in su ff ic ie ntl y se ns iti ve to t he po ssi ble salience of bac kg round informa tion, and she cha ng es her answe r to the ca nonical

PAGE 209

200 qu e sti on de pe nd ing on wh a t st ipu la te d b a c kg ro un d in fo rm a tio n s he is g ive n, it l oo ks a s if it is poss ible that the answe r to the ca nonical question that would ser ve as g ood evidenc e fo r a ph ilo so ph ic a l po sit ion is, " I do n' t kn ow ," or , in oth e r w or ds , it is e pis te mic a lly indeterminate . We can e x plain why we do not succ essfully perf orm TE2 by pointing out that the participa nt is not sensit ive to the re levanc e of ba ckg round informa tion to answering the canonic al question: we a re insuff iciently sensitive to the rele vance of bac kg round information bec ause w e have not been e x posed to a va riety of zombie thought experiments that differ in stipul ated ba ckg round informa tion. I t seems that once a participa nt is presented w ith differe nt stipul ated ba ckg round informa tion and rec og nizes that she cha ng es her answe r re lative to the bac kg round informa tion, she may chang e her answe r to, " I don' t know." E ven if the kind of c onceptua l knowledg e that the pa rticipant has is conditional, she may fail to rec og nize a backg round assumption she may have. Suppose, for e x ample, the pa rticipant knows a pr iori the following conditionals to be true: (3) I f the thing that in the actua l world that is the eff ect of tissue damag e and like cause s is a proper ty , which is, in fac t, epiphenomena l, then the conc ept of pa in d o e s n o t r e f e r t o a p h ys i c a l p r o p e r t y. (4) I f the thing that in the actua l world that is the eff ect of tissue damag e and like cause s is a proper ty , which is, in fac t, neurolog ical, then the c oncept of pa in ref ers to a phy sical prope rty . S u p p o s e a l s o t h a t t h e p a r t i c i p a n t t a c i t l y a s s u m e s t h a t t h e t h i n g t h a t a c t u a l l y p l a ys t h e p ai n ro l e i s a s t at e t h at i s , i n fa ct , ep i p h en o m en al . T h i s as s u m p t i o n m ay v er y we l l go unnoticed by the par ticipant. I f we do not provide the pa rticipant with zombi e thoug ht experiments that contain diffe rent stipulated bac kg round informa tion, she will answer the canonic al question in acc ordanc e with her backg round assumption, which may not be

PAGE 210

201 true. As a result, the thoug ht experiment will not be succ essfully perf ormed. Someone who fully possesses the c oncept in question may easily make this kind of mistake: she cannot sa y acc urate ly about the zombie targ et sce nario whe ther the be ing must be conscious bec ause she is not sensiti ve to re levant bac kg round informa tion, and, even worse, she may be re ly ing on a false ba ckg round assumption in answer ing the canonic al question. We do not know whether pain is phy sical or not, and it see ms that when we p re s en t a p ar t i ci p an t wi t h d i ff er en t s t i p u l at ed b ac k gro u n d i n fo rm at i o n , s h e w i l l ch an ge he r a ns we r t o th e c a no nic a l qu e sti on . Si nc e it i s a g e nu ine po ssi bil ity tha t pa rt ic ipa nts in TE2 ar e re ly ing on a possibly false ba ckg round assumption, we c an ca ll into questi on the anti-phy sicalist result of it. What I have tr ied to do in this chapter is show that the answe r may be that it is epistemically indeterminate whether the being descr ibed in the ta rg e t sc e na ri o mu st b e c on sc iou s. I f t he a ns we r t o T E2 is t ha t it is e pis te mic a lly indeterminate whether the being descr ibed in the targ et sce nario is consc ious, then TE2 do e s n ot s uc c e e d in pr ov idi ng Cha lme rs (1 99 6) wi th e vid e nc e fo r h is a nti -p hy sic a lis t arg ument. I t does not look as if issues about the truth of phy sicalism are resolved by the result of the zombie thoug ht experiment as usually perf ormed. Adding stipulated backg round informa tion does not make the way we e lic it int uit ion s a bo ut t he zomb ie tho ug ht e xpe ri me nt a po ste ri or i. T ha t is , h a vin g so me rele vant empirica l information about the ac tual world whe n we do a thought e x periment does not mean w e ar e not doing conce ptual analy sis. There a re c ases in which w e ar e better a ble to answe r ca nonical questions when w e ar e g iven stipulated bac kg round inf or ma tio n. Suc c e ssf ul p e rf or ma nc e of tho ug ht e xpe ri me nts inv olv ing tho se c on c e pts re qu ir e s e mpi ri c a l in fo rm a tio n, e sp e c ia lly if we a re tr y ing to m ini mize e pis te mic

PAGE 211

202 indeterminate ness as a n answe r to a c anonica l question. The way we a rrive a t the answe r to t he c a no nic a l qu e sti on on c e we a re g ive n th e re le va nt e mpi ri c a l in fo rm a tio n is a pr ior i. Another r eason w hy we may have g otten things wr ong in the zombi e thoug ht experiment is that the science of the bra in is "pre -atomic theor y "like. That is to say , as fa r a s th e re le va nt b ra in s c ie nc e s g o th a t mi g ht r e ve a l th e tr ue na tur e of c on sc iou sn e ss, it may well be that we are in a stag e that it is like the preatomic theory stag e for water and other c hemical kinds. I f so, then we are not in a position even to think clearly about the underly ing phy sical states that we might wish to identify with conscious states. We ar e like people in 1750 who c annot enter tain the Twin Ear th scena rios beca use it require s so me ru dim e nta ry kn ow le dg e of c he mic a l th e or y . We mig ht, the n, no t e ve n b e a ble to c on c e ive of the ri g ht k ind of tho ug ht e xpe ri me nt. Th is, it m ig ht b e su g g e ste d, ma ke s it easie r to overlook that we have stipulated the ' zombi es' phy sical states in such a way that we g uara ntee that wha tever is releva nt is retained. Cl os ing Re m ar ks I n this diss erta tion, I defe nd a view of conce ptual analy sis and thought experiments that preser ves the tra ditional practice of doing philosophy , but which a c c omm od a te s so me of wh a t th e e mpi ri c a l sc ie nc e s h a ve ta ug ht u s a bo ut t he wo rl d th a t, arg uably , is releva nt to philos ophical theor ies. The historica l frame work of the ories of conce pts that I sketche d in the first cha pter of this dissertation illustrates the important chang es theorie s of conc epts have underg one, in par ticular a chang e in foc us from fi g ur ing ou t w ha t c on c e pts a re to f ig ur ing ou t w ha t it is t o p os se ss a c on c e pt. I the n u se the notion of conc ept possession to develop a f rame work for how we mig ht think about doing thoug ht experiments. This is important since the pr actice of doing thought experiments is one that is not usually the foc us of philosophical discussion, but, which I

PAGE 212

203 a rg ue , is imp or ta nt i n d e a lin g wi th i ssu e s a bo ut s ke pti c ism a bo ut t he pr a c tic e a nd in ma kin g su re tha t th e re su lts of tho ug ht e xpe ri me nts c a n b e us e d a s e vid e nc e fo r o r a g a ins t ph ilo so ph ic a l th e or ie s. I n li g ht o f m y fr a me wo rk fo r t hin kin g a bo ut t ho ug ht e xpe ri me nts , I addre ss some historical cha lleng es to doing conce ptual analy sis and thought experiments. Finally , I apply the fra mework I have de veloped to a c ontroversia l thought experiment and explain how we may have be en g etting the wrong result, thus making room for a position i n the philosophy of mind that many have thoug ht is ruled out by that tho ug ht e xpe ri me nt. Many philosophers have taken the zombie thoug ht experiment to provide a kn oc kdo wn a rg ume nt a g a ins t ph y sic a lis m. T his is n ot a pr ob le m f or ma ny ph y sic a lis ts, since many phy sicalists are ske ptical about using thought e x periments as e vidence for or a g a ins t ph ilo so ph ic a l th e or ie s. Ho we ve r, the zomb ie tho ug ht e xpe ri me nt d oe s p re se nt a pr ob le m to ph y sic a lis ts w ho a re no t sk e pti c a l a bo ut t ho ug ht e xpe ri me nts . M y explanation of what may be g oing on in the zombie thoug ht experiment makes room for the ph y sic a lis t w ho fa vo rs the us e of tho ug ht e xpe ri me nts . I t lo ok s a s if we ma y be a ble to support both phy sicalism and the tra ditional way of doing philosophy .

PAGE 213

204 REFERENCES Austin, J .L . 1956. A Plea for Ex cuses. I n Ph ilo so ph ic al P ap e rs , edited by J . O. Urmson a nd G. J. War no c k. Oxf or d: O xfor d U niv e rs ity Pr e ss. Be aler , Geor g e. 1987. Philosophical L imits of Scientific Esse ntialism. I n Philosophical Perspec tives , edited by J ames E. Tobe rlin. Atasca dero, CA: Ridg eview P u b l i s h i n g C o m p a n y. ———. 1992. The I ncoher ence of Empiricism. Ar ist ote lia n S oc ie ty Supplementary Volume 66:99-138. ———. 1996. A P rio ri Knowledg e and the Scope of Philosophy . Philosophical Studies 81:121-42. ———. 1998a. I ntuition and the Autonomy of Philosophy . I n Rethinking Intuition: The Ps y c ho log y of I ntu iti on an d I ts R ole in P hil os op hic al I nq uir y , e dit e d b y M. R. DePaul and W. Ramsey . L anham, MD: Rowman & L ittlefield Publi shers, I nc. ———. 1998b. A Theor y of Concepts a nd Concept Possession. I n Con c e pts , edited by E. V i l l a n u e v a . A t a s c a d e r o , C A : R i d g e v i e w P u b l i s h i n g C o m p a n y. ———. 2002a. Modal Epistemolog y and Rationalism. I n Con c e iv ab ili ty an d P os sib ili ty , e dit e d b y T. S. G e nd le r a nd J. Ha wt ho rn e . O xfor d: O xfor d U niv e rs ity Pr e ss. ———. 2002b. Modal Epistemology and the Rationalist Renaissanc e. I n Con c e iv ab ili ty an d P os sib ili ty , edited by T. S. Gendler a nd J . Hawthor ne. Oxford: Oxford Un ive rs ity Pr e ss. Bloc k, Ned. 1981. Psy cholog ism and Be haviorism. The Philosophical Review 90 (1):543. Bog omolny , Alexander. 2005. Pythagorean Theore m . 2 0 0 5 [c it e d 1 5 M a y 2 0 0 5 ]. Av a ila ble fr om http:// cut-theknot.com/py thag oras . Br own, J ames Rober t. 1991. The Laboratory of the Mind . L ondon: Routl edg e. Bur g e, Ty ler. 1979. I ndividualism and the Mental. Midwest Studies in Philosophy 4:73121.

PAGE 214

205 Caldwell, Chris K. 2005. Euclid's Proof of the Infinitude of Primes . 2004 [cited 15 May 2005]. Available from htt p:/ /w ww .u tm. e du /r e se a rc h/p ri me s/n ote s/p ro of s/i nf ini te /e uc lid s. htm l . Carnap, Rudolph. 1952. Mea ning Postulates. Philosophical Studies 3:65-73. ———. 1956. Me an ing an d Ne c e ssi ty : A Stu dy in S e ma nti c s a nd Mo da l Lo gic . 2nd ed. Chi c a g o: U niv e rs ity of Chi c a g o Pr e ss. Chalmers, Da vid. 1996. The Con sc iou s M ind : I n S e ar c h o f a F un da me nta l Th e or y . New Yo rk : O xfor d U niv e rs ity Pr e ss. Chu rc h, Al on zo. 19 46 . A No te on the " Pa ra do x of A na ly sis " ; T he " Pa ra do x of A na ly sis " Ag ain: A Reply ; Analy sis and I dentity : A Rejoinder; How c an Ana ly sis be I n f o r m a t i v e ? The J ou rn al o f Sy mb oli c Log ic 11 (4):132-133. Davidson, Dona ld. 1980. Ag ency . I n Es sa y s o n A c tio ns an d E v e nts . Oxford: Ox ford Un ive rs ity Pr e ss. Fr eg e, Gottlob. 1882. L etter to Mar ty , 29.8.1882. I n The Frege R eader ( 1997) , edited by M. Be aney . Oxford: Blac kwell Publishers I nc. ———. 1891. Func tion and Concept. I n The Frege R eader ( 1997) , e d i t e d b y M . B e a n e y. Oxford: Blac kwell Publishers I nc. ———. 1892a. Comments on Sinn and Bede utung . I n The Frege R eader ( 1997) , edited by M. Be aney . Oxford: Blac kwell Publishers I nc. ———. 1892b. On Concept a nd Object. I n The Frege R eader ( 1997) , e dit e d b y M. Be aney . Oxford: Blac kwell Publishers I nc. ———. 1892c. On Sinn and Bede utung . I n The Frege R eader ( 1997) , e dit e d b y M. Be aney . Oxford: Blac kwell Publishers I nc. ———. 1893. Gr un dg e se tze de r A rit hm e tik , Volume I I . I n The Frege R eader ( 1997) , edited by M. Be aney . Oxford: Blac kwell Publishers, I nc. ———. 1918. Thoug ht. I n The Frege R eader ( 1997) , edited by M. Be aney . Oxford: Bla ckwe ll Publ ishers I nc. Gale, Richa rd. 1991. On Some Pernicious Thoug ht Ex periments. I n Tho ug ht E x pe rim e nts in Science and Philosophy , edited by T. Horowitz and G. J. Massey . Savag e, MD: Rowman & L ittlefield Publi shers, I nc. Galilei, Galileo. 1638/1914. Dialogues Concerning the Two New Science s . Transla ted by H. Crew a nd A. de Salvio. New York: Dove r.

PAGE 215

206 Gea ch, Peter . 1956. Me nta l A c ts . L on do n: R ou tle dg e & Ke g a n Pa ul. Ge nd le r, Ta ma r S zab o. 19 98 . E xce pti on a l Pe rs on s: O n th e L imi ts o f I ma g ina ry Ca se s. Journal of Consciousness Studies 5 (5-6) :592-610. ———. 2000. Thought Experiment: O n the Powers and Limits of Imaginary Cases . New York: Ga rland Publishing, I nc. G e t t i e r , E d m u n d . 1 9 6 3 . I s J u s t i f i e d T r u e B e l i e f K n o w l e d g e ? An aly sis 23:121-23. Gorman, Pete r. 1979. Py tha go ra s: A Li fe . L ondon: Routl edg e & Keg an Paul L td. Grice , H.P. 1961. The Causal The ory of Perc eption (1988). I n Perce ptual Know ledge , e dit e d b y J. Da nc y . O xfor d: O xfor d U niv e rs ity Pr e ss. Hig htower, Sandr a, Troy J ones, Eddie Ke el, and Vir g inia Urba n. 2005. A Brief History of the Pythagorean Theore m . 1997 [cited 15 May 2005]. Available from htt p:/ /w ww .g e om. uiu c .e du /~ de mo5 33 7/G ro up 3/h ist .h tml . Hi ll, Chr ist op he r S ., a nd B ri a n P. Mc L a ug hli n. 19 99 . T he re Ar e F e we r T hin g s in Re a lit y Than Ar e Dr eamt of in Chalmer s' s Phil osophy . Philosophy and Phenome nological Researc h L I X (2):445454. J ackson, F rank. 1998a . Fr om Me tap hy sic s to Et hic s: A D e fe ns e of C on c e ptu al A na ly sis . Oxf or d: C la re nd on Pr e ss. ———. 1998b. What Mary Didn' t Know. I n The Nature of Consciousness: Philosophical Debates ( 1998) , edited by N. B lock, O. F lanag an and G . Guzeldere . Cambridg e, MA : T he MI T Pr e ss. Kahn, Char les, H. 2001. Py tha go ra s a nd the Py tha go re an s: A B rie f H ist or y . I ndianapolis: Hac kett Publis hing Company , I nc. Ka nu c k, Ka thr y n. 20 02 . G e nd le r o n Wh y We C a n' t T ru st T ho ug ht E xpe ri me nts in Pe rs on a l I de nti ty . Pa pe r r e a d a t Pa pe rs of the 25 th I nte rn a tio na l Witt g e ns te in Sy mposium: Persons. An I nterdisciplinary Approa ch, Aug ust 11-17, at Kirchsbe rg am Wechsel, Austria . Kanuc k, Kathry n T. 2000. Personal I dentity : The Signific ance of the F ission Thought Ex periment. Maste r of Ar ts, Phi losophy , University of F lorida, Ga inesville. Kripke, Saul. 1980. Nam ing an d Ne c e ssi ty . Ca mbr idg e , M A: Ha rv a rd Un ive rs ity Pr e ss. L ang ford, C. L . 1942. The Notion of Ana ly sis in Moore' s Phil osophy . I n The Philosophy of G .E . M oo re , e dit e d b y P. A . Sc hil pp . L a Sa lle , I L : O pe n Co ur t.

PAGE 216

207 L evine, Joseph. 2001. Pu rp le Ha ze : T he Pu zzl e of C on sc iou sn e ss . Oxford: Ox ford Un ive rs ity Pr e ss. L oar, B rian. 1990. Phenomena l States. I n The Nature of Consciousness (1997) , edited by N. B loc k, O. F la na g a n a nd G. Gu zel de re . Ca mbr idg e , M A: Th e MI T Pr e ss. L udwig , Kirk, and Ray , Gre g . 2002. Vag ueness a nd the Sorites Paradox. I n Language and Mind , e dit e d b y J. To mbe rl in. At a sc a de ro , CA : Rid g e vie w Pr e ss. McGinn, Colin. 1998. Can W e Solve the Mind-B ody Problem? I n The Nature of Con sc iou sn e ss , edited by N. B lock, Owe n Flana g an, and G uven Guzelder e. Ca mbr idg e , M A: Th e MI T Pr e ss. Moore, G.E . 1942. A Reply to My Critics. I n The Ph ilo so ph y of G .E . M oo re , e dit e d b y P. A. Sc hil pp . L a Sa lle , I L : O pe n Co ur t. Nag el, Thomas. 1974. What I s I t L ike to Be a B at? I n The Nat ur e of C on sc iou sn e ss: Philosophical Debates ( 1997) , edited by N. B lock, O. F lanag an and G . Guzeldere . Ca mbr idg e , M A: Th e MI T Pr e ss. Neug ebaue r, O. 1970. The Ex ac t Sc ie nc e s in An tiq uit y . Pr ov ide nc e : B ro wn Un ive rs ity Pr e ss. P e a c o c k e , C h r i s t o p h e r . 1 9 8 9 . W h a t a r e C o n c e p t s ? Midwest Studies in Philosophy XI V:128. ———. 1992. A S tud y of C on c e pts . Ca mbr idg e , M A: Th e MI T Pr e ss. Plato. Meno. I n Plato: Complete W orks ( 1997) . Transla ted by G. M. A.Grube and edited by J . M. Cooper and D . S. Hutchinson. I ndianapolis: Hac kett Publis hing C o m p a n y. Pust, J oel. 2000. Intuitions as Evidence . New Y ork: Gar land Publishi ng Company , I nc. Putnam, Hilary . 1975. The Mea ning of Meaning . I n Philosophical Papers, Vol. 2: Mind, Lan gu ag e an d R e ali ty . Ca mbr idg e : Ca mbr idg e Un ive rs ity Pr e ss. Quine, W.V.O. 1949. Truth B y Convention. I n Re ad ing s in Ph ilo so ph ic al A na ly sis , e dit e d b y H. F . a . W. Se lle rs . N e w Y or k: A pp le ton -C e ntu ry -C ro ft s. ———. 1960. W ord and Object . Ca mbr idg e , M A: MI T Pr e ss. ———. 1980. Two Dog mas of Empiricism. I n From a Logical Point of View . Ca mbr idg e , M A: Ha rv a rd Un ive rs ity Pr e ss. Ry le, Gilbert. 1930. Ar e The re Propositions? I n Collected Papers ( 1971) . New Y ork: Ba rnes a nd Noble.

PAGE 217

208 ———. 1932. Sy stematically Misleading Ex pressions. I n Collected Papers ( 1971) . New York: B arne s and Noble. ———. 1945. Philos ophical Ar g uments. I n Collected Papers ( 1971) . New Y ork: Ba rnes and Noble. ———. 1946. Knowing How and K nowing That. I n Collected Papers ( 1971) . New York: B arne s and Noble. ———. 1953. Ordinar y L ang uag e. I n Collected Papers ( 1971) . New Y ork: Ba rnes a nd Noble. ———. 1957. The The ory of Mea ning. I n Collected Papers ( 1971) . New Y ork: Ba rnes and Noble. ———. 1962a. Phenomenolog y Ver sus "The Concept of Mind" . I n Col le c te d P ap e rs ( 1971) . New Y ork: Ba rnes a nd Noble. ———. 1962b. Thinking Thoug hts and Having Concepts. I n Collected Papers ( 1971) . New Y ork: Ba rnes a nd Noble. ———. 1984. The Concept of Mind . Ch ic a g o: T he Un ive rs ity of Chi c a g o Pr e ss. Sellars, Winifred. 1997. Empiricism and the Philosophy of Mind . Cambridg e, MA: Ha rv a rd Un ive rs ity Pr e ss. Smart, J .J .C. 1959. Sensations and Br ain Proce sses. Philosophical Review 68:141-56. Sos a , E rn e st. 19 98 . Re thi nk ing I ntu iti on : T he Psy c ho log y of I ntu iti on a nd I ts R ole in Philos ophical I nquiry . I n Rethinking Intuition , e d it e d b y M . R. D e Pa u l a n d W. Ramsey . L anham, MD: Rowman & L ittlefield Publi shers, I nc. Stanley , J ., and T. Willi amson. 2001. Knowing How. Journal of Philosophy 98 (8):41144. Wil kes, Ka thleen. 1988. Re al P e op le : P e rso na l I de nti ty W ith ou t Th ou gh t E x pe rim e nts . Oxf or d: C la re nd on Pr e ss. Wil liams, Ber nard. 1970. The Self and the F uture. I n Personal Identity ( 1975) , edited by J. Pe rr y . B e rk e le y : U niv e rs ity of Ca lif or nia Pr e ss. Wit mer, D. G ene. 2001. Conc eptual Ana ly sis, Ci rcula rity , and the Commitments of Phy sicalism. Acta Analytica 16:119-133. Wit tge nstein, L udwig . 1958. Philosophical Investigations . T ra ns la te d b y G. E. M. Anscombe. 3r d ed. Eng lewood Cliffs, NJ: Ba sil Blackw ell & Mott, L td.

PAGE 218

209 ———. [1921] 1997. Tractatus L ogico-Philosophicus . T rans lat ed b y D. F. P ears and B. F. Mc Guinness. L ondon: Routl edg e. Yablo, Stephen. 2000. Te x tbook Kripkea nism and the Open Te x ture of Conc epts. Pa c ifi c Ph ilo so ph ic al Q ua rte rly 81:98-122.

PAGE 219

210 B I OG RA PHI CA L SKE TCH Kathry n L ammens wa s born in For est Hills, New Yor k in 1975. She rec eived he r bache lor' s deg ree with honors from Hof stra Univer sity in Hempstead, N ew Yor k, wher e she double majore d in philosophy and Eng lish. S he re ceive d both her master ' s deg ree a nd do c tor a te de g re e in p hil os op hy a t th e Un ive rs ity of F lor ida . Sh e is m a rr ie d to Phi lip L ammens, and the y have two c hildren, Emma and Ma tthew. Afte r g radua ting f rom the University of F lorida, Ka thry n and her family are moving to Virg inia wher e she w ill be a V i s i t i n g A s s i s t a n t P r o f e s s o r a t J a m e s M a d i s o n U n i v e r s i t y.