Jonathan W Schooler & Charles A. Schreiber

Exp erience, M eta- cons ciousnes s,and the Paradox of Introspection

Abstfact: Introspection is paftdoxical in that it is simultaneously so compcllingyet so elusive, This paradox emerges because ahhough expeicnce itselfis indis-putablc, our ability to explicitly characterize etpe ence is ofren inadequate.Ultimately, the accurucy of introspectiw rcports d?pends on indiyi(lu.lls' imper.fect ahility to takc stock (i.e., to become tfieta-conscious) (t theb expcrience.Although there is no ideal yerdstickfor assessing introspection, examinatioh ofthe degrce to which seu-rcports syrtethatically .ovaty with thc cnvironmental,behal)ioural, andphrsioloeiel concoftitants ofetperience can help to establishthe correspondcnce between meta"consciousness and experience, We illu:trutethe viability (t,tuch an approach in three domains, imagery, mind-yunderihg,and hcdonic apprcisaL idefining both the situations in which intrcsnectionsappear to be accurdte and those in which they seem to diverye.from underlyingexperience. We conclude with a aliscussion of thc various.factors (includingissues of detection, trunsformation, and substitution) that may cause met1-con-sciousnes! to misreprerent experience.

The paradox ofintrospection stems from the fact that personal experience cone-sponds to that which we know best subjectively, yet least empirically. Subjec-tively speaking, there is nothing we know better than our own introspectiveexperience. As James (1890) observed 'lntrospective observation is what wehave to rely on first and foremost and always' (p.185). Indeed, as Descartes(1637) noted long ago, nothing seems to offer more ontological certainty thanintrospection, as nothing is more incontrovertible than one's knowledge of par-taking in experience. lfyou can inffospect, then you may be dreaming, you mightbe abrain in a vat, but there seems absolutely no question but that you are havingan experience.

Allhough subiectively incontrovertible, it is impossible to directly assess thecontents ofexperience, and thus no decisive way to empirically determine when

Jonathan W. Schooler, Depadmenl of Psychology, Unrversity of Bntish Cotumbia, 2.136 Westl',4all, Vancouver. BC. Canada V6T 124.

Journal oICon!. iousnesr S/edier, l l ,No.7 8,2004, pp. j7 39

lli J.W SCHOOLI]R & C.A. SCHREIBER

reportcd introspections accurately vs inaccuratcly characterizc underlyingexpcrience. Thc empiric elusivencss of subjcctive experiencc is powerfullyi l lustrated by Dcnnett 's (199I ) c laim th.r t qual ia i tsel f is an i l lusion- Dennett 'sargumcnt builds on the obscrvation that the f:ct thal wc believe oursclves b possess qualia does not in itself provide evidence that wc necessarily do Zombies(philosophic.rl constructs who apPear and behave exsctly like humans but wholack internalexpericoce) would also bel ievcand vehemently assentha( thcy pos-se..eLl r i r id \uhtecl i \c e\pericncc. Dennctr lunh(r rrguc\ lhr l becau'c i l i 'impossiblc to ful ly pin down what any part icular expcricnce is real ly l ;ke' theconstruct of subjective cxPerience is too vaSuc !o be meuningful (For example'whcn ;ndividuals we,rr invcned ! l lasscs:r trd. af lcr a whi le begin to navigalc suc-cesstul ly in thc world, is lhcir subjecl ive expericncc reve|ted or do they simplylcrrn to adiust to an upside down wodd?) Given thc infcasibi l i ty of c i ther demonstrLr{ ing the existcnce ol subjcct ive cxperience or charlcter iz ing whrl i1 isactua' ly l ike, Denoctt coDcludcs, 'nobody is c(rnscioLts .- . we are al l Tombies' . '

The str ik ing thing about Dcnnctt s cr i t ique of subjccl ivc expericncc is thatI tom an ernpir icr l pelspcct ive hc is ahsoltr lc ly r iSht Whcn rcly ing on thc rules olevidence upon which scicnce is bascd ( i .c. cmpir ic l indcPendent rcportableobserval ions) wc sinrply cannot know thi t t we arc consci{)us al al l Neverlhclcss.(and this is whcrc wc divcrgc f f t )m Dcnn€t t ) wc do. l t seems self 'cvident lo us(and wc cxDecl lo nl{ny rcaders) thi l l the exislcrrcc of subjcct ivc expcricncc'thouSh eDtirc ly eluding the sources ol-cmpir ical cvidence upon which al l othcrknowlerJgc is hLrscd, ncvcr lholcss remuins i tn i l ld ispulablc tacl whosc certuinly isa( lcasl (nr pirr with, rn( l nr! ,uably cxceeds, anylhiDg clsc that we know (tbr s imi-lar argunronts scc Chit l rncrs. 1996; Dcsci l r tcs, I6371 Sel lars, 1963)

Civcn the Daradox of introspcct ion ( ic , thr i expcricncc is subjcot ively\cl f evidcnl hut ernpir ic l ly i l )scrutt lb lc). i t might sccm templ ing to simply l lb l ln_don any ir l tcrrpts i r l cmpir ical invcst ignl ions ol subicct ive expcicnce lDdccd.this strr legy chrr i |ctc. izcs the bulk 01 psychologic l rcscxrch ol thc 1l |s1ceDtuly 'Howcvcf. i1 is our view thl t t by conf(rnt ing thc PiLr dox of inlrospect iot l hei ld on'\ c. . l l r rr . rkc h. i rJwJ). s l ,ccr l i . : l l l ] . t l r rcc imf, ' f l in l (rrn\ i i j { r i r l r ' rn ' erncrSc Irrrrnthc observat ion that expericnce is ai once subicct ively scl f :evidcnt l rndempir ical ly inscrut blc-

Epistemctogically speakinS, the paradox ofintrospeclion tclls us that there is noway tha( wc will evcr be.rblc to empirically know subjectivc cxpericncc in themanncr in which we cmPir ical ly know other $ings. Unl;ke other elusive coocefts(c.g., distrnt galaxics. tiDy pllnicles) whose cmPirical invcsti!iirtion cvades ltnalysis due 10 limitalions ofmeasuremenl, subjectivc cxperiencc is. as Dennctl s ilnalysis iltu\lratcs, empirically nol) demonstrable- Th;s inscrulability mcans thrlcmpidcal investigaiions of subjeclivc exPe{icnce mus( nccessaily bc groundcd inhumility, cogDizrnt th t any accounl is oPen to alternativc interyrelatiors

Altcr duking thescellinrs. Dcnncll irggcsts lhdt il {ould be rn acl ol desperatc intelleclu'ldLshon

",turoouot"rt'isrxc,ri,u'ouiotcontcxt lP 106) Rcdcrs arc rhcrcl're duliiulll ad viscd tl) rc!'ew

$c con;cxl iD vhich lhc\e rss.(ons re m!.lc and fom theil own oPini(nr aboul whcthe' wc hrle

THF P \R AI) ' I \ OF INTROSPI:CTION t,

Methodologicr l ly spel* ing. thc paradox of inrrospect ion i l lustf t r tes hor, ,invesl igat ion of subject ive expericncc also af i i ) rds a lool uniquc to o(her modcsofinquiry. Specifically, we can usc thc fac( that *,e havc experiencc to evaluatc(he models ofcxpcrience that wc dcvclop. The seeking of r concordxnce ofihcconclusions of cmpir ical invest igat ions with the f i rs l person pcrspect ives ofthose l tenerat ing and ev.r luat ing theories represcnls a central element oIphenomcnologicul psychobgy (c.9., Varela, 1996, bur see Schooler & Dougal,I999). And al though perhaps rarcly cxpl ic i t ly acknowledged. in al l donlt ins otpsychology .escarchers' wi i l ingncss to rccepl theoret ical coDclusioDs is al lectcdby whethcr lhcy l lnd such conclusions int fospcci ivcly compcl l ing. (X coursc.this incvi trbly merns thar somc subject iv i ty is introduced inlo the sciencc ofintrospcct ion. But, ul ter al l . isn'1 thul the point?

Final ly. l .om a thcorcl ical perspecrive, thc prradox of introspect ion hight ighlsadisl inct ion thrt has bccD i tnpl ic i t thfoughoul lhc foreSoing discussi0n: namcly,that bclwccn the xctual cdr lcnls ol an expcf icncc (cxp€r ient i f l l consciousncss)and one s expl ic i t hcl icfs bout rhecontcnts ol-consciousncss, here intefchangcably rclerred to.rs mcla-consciousness (Schoolcr. 2{XX); 200211) or mcta-arvarcrcss ( i lck & Shal l icc.200l l Schoolcr.2(X)l) . t inrpir ic.r l ly spcakinB. whcn wcsk part ic ipanls lo int()spect. r l l Ih l l r wc can gamer is whal they i r fc nrcl i l ,con

scioos of ( i .c- , whal lhoy /r , I i . , r ' { , thcy i t ro cxporiencin8). Froot such nrct i l -conscious reporls wc rrust then t tenrpl 1o inl t r whi i t they ar.c :rctui t l ly cxpcf icncing.lund i t is lhc possibi l i ly ol dissocir i t ions bctwcsn expcricut i r l consciousncss undtnct i r i , , r r . , : i , ' r rsI , .s, thrt p, .( \ rh. . pr, ,hl( r

Al(housh lhc disl incl i (Jn hclwccn cxporiont i r l consciousness nd rDctr-consciousrrcss does n( ' l c l i rnin lc lhc pruirdr lx of i r r l | \ )spccl ion. i t d()cs help u\ t r) scchow l l subiecl ivcly iDl i )rmc(lcrnpir ic lscicncc ol in lrospect ion nr ight pn)cccd.On thc one hand. thc dist inct ioD wheD considcrcd in l ight of thc l : rrrdox ol iot f t )specl ion. i l lustrr tcs how sel l - fepor(s, wl l ic l l [cccss r i ly nrLrst bc bAsed or tnclconscious rpprr is l ls ol experience. cl ln bc f l wcd. even i l (hcy iLfe birscd oexpcriencc. whish is i lsel f sel f cvidcn(. On rh€ o lhef hrud. i r suggcsts {hrr i t 'cnlpir icalobservat ions r iJgarding when sel l -reportscoincide with othcrpft)bableindicators of expericDcc arc considefed in the contcxr ol humil i ty and inl i ) rmedl i .st-person pcrspcct ivcs, lheD i t rr ly be possiblc 1o scicnt i f ic. l l ly advanccprinci f lcd conjectures l rbout thc n. i turc of expenence.

In thc fol lowing anrlysis. wc futuhcr f lcsh out the naturc of such an appro,tchdrawing oD recent considcrrt iots of how a dist incr i(rr bctweeD expcricncc andmela-arvarencss can help l(J resolve sonrc of thc inherenl difticuhics as$ciaredwith introspcct ion.Welheni l lustratethcvalucof(hisapproachinthreedonrainswhcrc iolrospect ion plays a pirr t icular ly cruci l role: i r ragcry, rnind- wandering,

lmplications of Dissoriations lJetwe€n Expcricnc€and Meta-Awarcncss for Introspection

Sorne ol the eaf l icst obscrvat ions reg rding the chal lcnges of inrrospccrivcl)re 'descf ibing cxpcricncc cdrne, not \ufPris ingly. f iorn Wil l i l t ln lancs ( I891))

20 J,W. SCIIOOI-I ]R & C.A. SCHRBII]ER

who obscrved thar( 'We f ind oursclvcs in cont inual error ancl uncerlr inty so soonas wc iuc cal led on to D1tme and class and not merely f ic l ' (p. l9 l) . J lmes comparcd attcrnpts l(J rellectively pin down flcctinS subjectivc cxperiences k) that otseiziDg ' . r spinning rcp to calch i ls motion, or t fy ing lo lurn up the gas quicklyenough to see how the darkncss looks (p. 244). Jamcs recognized that thc challenge of introspect ivc fc-dcscr ipt ion stems in pan from the l imit l t ions of lan!tuage, not ing thai 'abscncc of special vocabulary for subject ive facls hinde$the study of al l hutthc vcry co,rrsestof them' (p. 195).

As a conscqucncc of thc di f f icul t ies of int fospcct ion recognized by James .rDdmorc vchcmcnl ly decr ied hy Watson ( l92l) and olheIs. thc usc of introspect ivccvidcnce within n instrern psychology was hrgely:rvoidcd thrcughout muchof thc pasl cen(ury. Ho*'evcf. an inrportaDt breLrkrhrough in the acceptancc ofsel i r repor ldataf i ) l lowcdl , l r icssonlndSimon's( l980:1993)serninalworkdocunient iDlt the many condi l ions undcr which think-aloud prorocols (where panic-ipanls si f ip ly statc whatcvcr thorghls cr(rss their mind as (hcy engage in a task)off t r val id dala tha( closcly corfespond to bchaviour without notably af lccl ingpcrtbrmancc. l ' l r icsson rnd Sirnon lur lhcf adiculatcd the c{nldir ions urdcr whichthink-aloud prclocols l rre l ikeiy to hc vr l id (c.g.. i f they Lrrc donc couctrrent lywithoul plr t is ipaD{s attcnrpt inS lo cxpl i t in the pf i )ccsscs undcrly ing thci llhoughls) aDd whcr thcy t l fc l ikcly to bc i rv l l i ( l or reacr ivc (c.g.. i f lhcy ore retro-spcc( ivc or requi le parl ic iprnls to nukc inlcrcmce\ Lthoul why rhcy i rrc bchavingin a parl iculrr wry).

Er icsson and Sirnon s considcr. t t ion ol thc possihlc rci lc l iv i ly ol think-aloudprotocols. r i rd lhcir drcuDrontrt i ( )n oj thc l icqLrcnt corrcspondl jnc(r hctwccn thccon{eDts ol iDdividuals ' (hinl l l loLrd protocols nd trsk pcrtbrnlancc, rcpresenlsaDirnpo|1 nt l i )undir l ion rpon which thc prcscnl ni l lysi \ is b. l icd. I lowcver, intheir eDlhu\ iasf i to rcestuhl ish lhc ! l id iry ofscl l -repo| ' l tcchniqucs,thcy nor.r-hly dc crrphrsizcd lhc p{}\s ibi l i ty th 1\,efbal ly fcpor '{ i0g onc s rhoughrs couklforcc individur ls to vcrb l ly re-repfcscnt inhcrcnt ly non-verbr l expcricnccs.Argultbly. in so doiDg, they ovedookc(l thc possibi l i ty ( nLi evidcncc) rhar disroft ions or omissions crn occur even when indiv iduals s i rnply th ink r loud(Schoolcr.r . / / . , l993tL ne&Schoolef. inpfess;thoughscchrics\on,2002t i) facr i t iquc of this pcNpecl ive).

Recert ly, thfee independent scts of invcsl igatoh of l l red conpcl l ing argument\ tbr how discrcprncics bctwceD thc conlents ol expe| iencc and individuals 'cxpl ic i l awarcncss of (hose coDteDls can lead to di f f icul t ics iD scl f rcpod datx(Jack & Shal l ice,2001:Jick & Rocpstorf f ,2002; Lrmbie & Marccl .2002;Schoolcr. 2000; 2001; 2002b). As Schooler (2(X)2b.) observcd: I f , as i rgued.mela consciousncss requires re- represent in! i the contents ofconsciousress, theni1 litlows that rs with rrny recoding prcccss. solltc iofoflD.rtion may becoDre lostof dist(nted in the l fanslat ion' (p. 342). Iu thc conlcx{ of in lefpfet ing fai lurcs tLrrccogrizc cmotions Lambic al ld Marccl s irni l r r ly observed: 'Somc oI (hcrep{rr led cases ol '1ack(temol ionexperiencc' . . .sccmtocxhibi t thefeaturesol 'u hat our frurnework would descr ibc as I i rck of al , rzrrcss ol f i rst order emotionphcoomcnolLrgy' (p. 249). Likcwisc J.rck rnd Shrl l ice pointed out that

THl i PARAI)OX Ol, IN I ROSPEC I ION 2l

'inlrospeclivc rcports nray bc sccn as thc product oftwo factors: firsily thc "rawdata" . . . and sccordly thc conceptual f ramework. or '1nodel". . . . The cxtcnt towhich subiccts arc conect depends on the v?r l id i ty oftheir model ' (p. 177).

These rescarchcrs also ovcrlappcd rcgarding the metric by which djssociations belwecn cxpcricncc and nreta-consciousness are ident;ficd, with cachenlphasizing thc importancc ofdissociat ions between self-repons and othcr indi-rect indices. Iror cxamplc. bolh Lambjc and Marcel (2002) aod Schoolcr (2000;2002b) suggcslcd that discrcpancics hetween fepresso.s' physiologicalresponses to slress and thcir scl f-repofts (e.g.. Asendorph & Schefcr, I9i l3) nraybeindicat iveofrepressors' lackofrwarcncssofthcirownsubicct ivcslalc.Si inilar ly, lack Lrnd Sh.r l l ico (2001) argucd that discrcpancics bctwccn chi ldrcD'sself ' repo. ls Lrnd their behaviour in working on algcbra problcr ls ls icgler &Srcrn, l99l l ) indicated that chi ldren ' l i rcked irwarcncss o[ their own discoveryand usc ofthe struteSy (p. lTt i ) .

The str ik ing overlnp betwcen those thrcc largcly iDdcpcndcDl rnalyscs ol-(hcsoufccs of intnrspect ive erf t)r suggests (hat thc dis( incl ion bclween oxpol icncoand mcla-aw.rcncss is Lrn idea whose t ime hrs cornc (Jack, 2(X)4). Nevertholoss.(hcrc r .c also sor c inpo[Lrnt di l lerences bclwecn (hosc rppft) lchcs. lhoughrrcognir ing rhc v lLrc ol idcrrt i ly i ' rg dissocit l { ions bet lveen selFrep(n1s andbchirv ioLir i r lnrcrsufcs,Jackl lndShal l iccs(2(I) l )pr i rnrrysolut ionl i ) r lcsr) l ! ingdiscrcpi lncics in inl(rspccl i {)n involves ' rcpl i lc ing or rel- in ing lhe su\ iocls 'Inodcl l i ) r undcr\ trndin8lhcir own nrcnrr l str tes' (p. I78). Whi lc wo concur wi lh.hck nd Shrl l icc s rppl) ch lo thc cxlcnt thut \ {c recol lnirc ( nd i r)doed hirvcl i )und) thir t onhrnccd oxpcrl isc in dornain c n r l low v(rrhal protocols lhir l l rcInofc inl i ) rnr i ( ivc Lrnd less rclet ivc (c.g., Mclchcr& Schoolcr. 19961 irr prcss),r t tcDrpls l l fc l in ing pi | | t ic iplnls introspcct ivc rrr f i lc ls in i rr c l lo l l 1o cnhrrrccthcir rccurlcy i \ r lso [ i le wi lh chl l l lcngcs (Schoolcr. 2002 ). l r p r t icul i r . lhcfcis r lwrys thc f isk thrt lhc iolr(x lucl ion ol Dcw mcDtnl undcrstandirSs nr. ty l tcfthc very cxpcricrcc laf l ic iplots are atteln iDg loelucid,r(e.

lnlnbie nnd Marccl s (2002) i rpproreh tr) explor ing discreprncies bc(wecDphcnonrcnal crnrsciousne\s nd rnc{a-consciousnes\ laf8cly f i )cu\cs or pi t thologicr l crscs in which individul t ls show .ru absc cc ofrwrrcncss ofcxpcricnccdemoLional s lalcs. ' Ihough considefrt ion ol thc i rnpl ical ions of inlrospccl ivc di l l icul t ies i i ) r indi ! idur ls wi lh palhologies i \ i portanl, this emphasis overlooks thcc$es in which dissociLl l ions bctwccn consciousness Dd nrel -consciousnessnuy hc obse|ved wi lh nornt l i r ld iv idur ls.

' Ihe.rpprorch lakcn by Schooler (Schoolcr,2000; 20024,b; Schoolcr.r 1l / . .2003; Schooler & Schlcibcr. in press), and exteDded here, focuses on dissocirt ions between coDsciousncss rnd lne( iL consci{)usDess thr l may be involvcd ootouly with fathologie l c ses but also wi lh morc everydly exi lmfles. In so doiDg.i l in(roduccs i rulnher ol methodologics tht l are paft icul i l r ly ] lu i ted l i ) r Llocu'rnenl ing dissocia( ions in everydlty contexts iocluding using cxpcricnce srmpl ingto catch people engagimg io couscious str tes (e.g., ur ind wrndering) bcforc thcycatcb thcnselves (e.g.. Schoolcr. t .1/ . . in pfess). . rnd iDvesLigt l ing the i lnprcl ofcncouraging metr-conscious rellcclion Lrn pcdi)nnance (e-9., Wi1\oD & Schoolcr,

22 .t.w. scHoor_r-rR & c.A. SCHREIBER

l99l). This approach also articulatcs two distinct types ofdissociations betweenconsciousness and meta-consciousncss: tcmporal dissociations in which con-sciousness occurs in thc abscncc of mcta-consciousness, and tr nslation dissoci-ations in which meta-consciousness misrepresents underlying experience.Finally, this appro.rch, at leastas explicated here, grounds dissociations betweeoconsciolrsness and mela-consciousness wilhin the larger context of the paradoxof introspection-

One critical concern raised by the pararlox of introspection is that although itmay seem rcasonableto interpret d issoc iations between behavioural rcsponscs andself-repon as evidence of expcriencc without corresponding meta-awareness,because we cannot dircctly mcasure expcricnce, such aconclusion must be viewedwith caution. This inescapablc fact means that we must approach this topic withhumility, ever open 1() rlternative accounts, At the same time, there are variousstrategies that we may be uble rc employ in determining when self-reports aremore vs. iess likely to correspond to underlying experience. These include:(l) Multiple convergences. The more all sources of evidence converge on the

same conclusion, the more conl-idence we can reasonably have in that con'clusion. Thus if individuals' self-reports not only correspond rc their bchav-iours, but also sensibly f luctufie with changes in the envircnmcnt andconsistcntly map onto changes in underlying physiological state, then wccan become incrcasingly coofident that individuals' meta-consciousappraisals oftheir expericncc do in fact corrcspond to its actual undcrlyingcontent,

(2) Mult iplc divergences. Conversely, when individuals'self:repofts oonsis-tently divcr8c frcm environmental l luctuations, behavioural responses. andphysiological statcs. we nust become increasingly concerned thut meta-con sc iou s appraisals arclailing to adcquatc ly capturc undcrlyinS cxpcricncc.

(3) Reactivity ol sel l-reporl. I feDcouraging individuals to introspccr about thcirexpcrience systematiclrlly rlters. and particularly if i1 impairs, perfonnance,then it seems rcasonablc to spcculate that the intK)spections nray be tailinBto dojustice !o lhe undcrlying cxpcricnce.

(4) Appeals !o introspection. when dissociations bctwecn self-reporl and othcrpossible indiccs of experience are observed, there are always two possibleaccounts: (a) thc sclf reports misrepresented actual experiences, or (b) thealternative measures do not actually correspond io experience, AlthouBhresolution of this dilemma must be a malter of conjeclure, educated intro'spections regarding what sccms more consistent with tlrst person experi-ence may be useful for adjudicating between these alrernatives.

In the folhwing discussion, wc i l lustrate the viabil i ty ofthe above approach inthrec domains: imagery, mind-wandering, and hedonic appraisal.

Visual IrnageryFor many ycars psychologists squabbled about what to makc of thc commonintrospection that thc visual inrages in our'mind's eye' share similarities wifh

I Hh PARADOX OF ]NTROSPECTION 2l

actual visu.rl expcricnccs. Some researchcrs argued that these introspectionsreflected rnaloguc rcpfesentati{}ns lhal corrcs nrded to visual percepls (c.g..Kosslyn, l9t j0; Shcpafd & clDper, 1982). Others, however, insisted lhat v isualimagery w1|.\ no diffcrcni from symbolic propositions, and thus its uniquc visualUopert ies were i l lusory (e.9., Anderson, 1978; Plyshyn. I98l) . Al lhough tbcdebate was hald fought. ul t im.rtely (hosc arguing f i ) r the uniqueness of v isualimagcry caff ied (he day as evidenced by a rccent survey of psychologists(Rcisbefg er d1.,2002). At the same t imc, howcver. these methods havc alsorcvealed inlpor lanl domains in which introspcct ions about im{gery can bc lnis-lcading. We review thcsc lwo sides in turn.

I,ridcnk Ji)r the ( hunti(itt ol innger\ illtrotpcctkrnsA nu|nberofdi f fcrcnt paradigms hrvc bccn used todof i lonslralc lhcclose coffesln)ndence bo(wccn the n ture of inlrospcct ive im gcly rcports and ihc physicalprope ies ol lhc imaged sl i rnulus, as wel l rs the bchavioLrrnl and physiologicalrcsponses 10 thc imal ie. Sonle of lhc f i rst cviLience lor thc inagcry/pcfcept ionl inl came l iorn studics denronstrr t i r rg lhc correspondcltcc bclwccn thc ln niptrlal ion of v isui | l i rn:rgcs anLi physici i l pcfcepts 10 which (hosc imagcs corre\po ( lcd. F(n exrmplc, fcscirch on nronl l l rotat ion involving rsscssnrcnt ofthcnr:r tch hctwcen lwl) objccts. one ol which rust he nrental ly rol l l lcd i rr ordcr t() bccorrprrrcd wi(h lhc rnhcr, rcvcLrlr a vcry prccisc rclr l ionshiP bolwccn lhc rrn(ru!11{r l t iDrc i r la lcs (( ' r l i rkc r i . iudSnrcnt und lhc urrr(nrr) t ol Physicir l rol l i ( )n i l wonldreqLri fc1orl ign lhe i tcrns(Shcprrrd & Mclr lcr. l97l) . Sir i lar ly, thcl inrc i l l . rkcsto scrn l ror l r ()nc locir l i (nr l ( ) rr)olhcr on i r v isLr l izcdnrr lPishighl)conclnlc( lwi lhl l rc rclunl dis l l lnccs helwoorr lhosc l { lcrr l ions (Kosslyn. l9 l l3) Behrvir tural cvi( lcrrec l ()r thc closc eoffesfondcucc hclwcen inr lgcry and vision crtnre l i r tn slut lics ( lcr {nrsl i r l i | lg thr l v i \uul xDd vcfhr l s(f l tegics Pnxluccd dist incl pir t torns olrcl lc( ion t i rncs.rs\ocir i lc( l rv i(h ver i ly inA lhc rclr l ionshiP bclwccl l pic lurcs (e ! , . . r rphysicr l+rbo\,o physic i l ' i ) i rndscnlcnccs(e.g. ,1 l lL l ' j l -L lSISABOVITt l l iSTAR) ( Mrtcleol l u /r / / . . l t )7l t ) .Neurouogri l ivcevidencelbrthcclosccorre\pon-dcncc betwecn vision rnd inaScry wxs pfovidcd hy the numeft)us studics dcmonsrr i r t ing thir l cnSrging in visur l inr l l iery ct ivatcs br l l in fegions associalcdwi lh vis ioD (e.g.. Famh. l995r,b).

Et,idenct lbr luilurcv oJ ittlto\rtctions illrolring itt.ltgt'nAhhough ihe gcncfr l in l f t )spccl ion thi l t imagery is l ikc vis ion is norv strolSlysupporlcd. there erc also sitLrutions whcrc intft)speclions itbout ilniLgery itppcar tofai l . I ior example. in a lel lcr imrging 1. lsk Kosslyn. Cave, Provost, aDd vonClicfke (1988) found th lal though part ic ipaDls bcl ievethat let ters arc imaged.r l1rt oncc. their rcrcl ion t imes indicalcd that tncstal imrges rre conslructcd in i tpiecerncal f lshior. Sirr i lxr ly. Clh.rmhcrs and Rcisbcrg (1992) lound thi ll lhough paft ic iprnts bcl icve thl t l thcir images arc cqual ly dense al l over, thcir

rbi l i ty to usc imrl iefy to orrke detenl l inat ions about charactef ist ics ol , r prcviously cncoded rcvcft ib le l igurc (duck/fabbit) depends on which palr ol the

J.W. SCHOOLER & C.A. SCHREIBER

image they viewed as corresponding to the face. This finding suggests that con-trary to introspection, image clarity varies as a function of the direction ofattention (Reisberg, 1996)-

Another source of error in imagery inffospection comes from verbal descrip-tion. A substantial body of ev idence reveal s that when individuals verbally intro-spect(i.e., attempt to describe in great detail) complex non_verbal images such asthe appearance of a previously seen face, disruption can ensue- For example'Schooler and Engstler-Schooler (1990) found that participants who described indetail the appearance of a pleviously seen face were markedly worse than con-trols on a subsequedt recognition test. Additional studies demonstrated that thisinterference, termed 'verbal overshadowing', Senelalized to another non_verbalstimulus (colours) but not to a verbalizable one (a spoken statement). Moreovel,the type of introspection was found to be critical. Whereas verbal descriptionwas disruptive, visual introspection (i.e., just imagining the face) had no effect.since its original demonstration, verbal overshadowing has been replicated innumerous laboratories, albeit not always (for a meta_enalysis see Meissner'2002). lt has also been lound to genemlize to other visual stimuli includingnon-verbal forms (Brandimonte er di., 1997), maps (Fiorc & Schooler,2002),and the appearance of mush rooms (Melcher& Schooler, in press), and alsoothermodalities such as audition, including both music (Houser at 4r., 2003) andvoi.es (Pefiect et a\.,2002), as well as taste (Melcher & Schooler, 1996). Suchdisruption suggests that verbal introspection fails to adequately capture ineffableexperiences, breaking them apart in a manner that makes it difficult to put backtogethet (see Schooler et al., 199'l, Schooler, 2002c for additional theoreticaldiscussions of this effect)

Summary and caveatsExaminalion of the relationship between imagery and introspection illustratesboth the successes and the limitations ofinlrospection. Although once called intosedous question, the converging evidence demonstrating thc ck)se correspon-dence between images and external stimuli, behaviours, and brain activationhave consislently supported the reality of this experience. Despitc such evi-dence, the empirical inscrutability ofexperience leaves opcn the possibility thatthe correspondences between imagery and perception occur because both areexperienced exclusively symbolically, While such an account cannot be ruledout, it is here whete the potential upside of the Paradox of introspcction may beuseful in suggesting an appeal to readers' first-person experience How plausibleis it that your sensory experience of a red apPle is purely symbolic? And if not'given the similarities between your experience ofan apple and your image of anapple, could an image ofan apple be exclusively symbolic? While clearly cogni-zant of the pitfalls of appealing to the intiospective observatjons of thc reader'we nevertheless suggest thlrt the confluence of behavioural, physiological, andself-rcport dafa, when combined with thc first-person perspectlve, makes astrong case fbr the reality of imagery

THE PARADOX OF INTROSPI1C I ION 25

Whilc the pfesent rpproach illustrrlcs thc succcsscs of introspcction forreveal ing ced?rin specls o[ imagcry cxpcricnccs. i t a lso highl ights i ts di fTicul t iesin other fespects. Specif ical ly, discrcpancics between individuals ' sel f : repoftedintrospcctions and their behaviours on a varicty of measures ;ndicatc that intro-spcctior s are not alwry s xccu r lc. Morco vcrt rcact ivc effects of vcrbal i ntrospec-tions suggest th.rt this is pa(icularly thc case when individuals attenrpt t()rranslalc ihcir non-verbal images i l1to words. l lowcvcr. oocc agai l somc humil-i ly is rcquired in interpret ing such indings. For cxamplc, in charactcr iz ing thecffects of vefb?rl1y describinS rnclnory for p,cviously encountercd non-verbtlst i rnul i . lhc rc-dcscr ipt ion process rnighl nl(emativcly bc characlcr izcd l rs simplya descrip( ion of thc st inulus. J ck and Rocpslorf l (2002b) proposc such anaccount in rrguing that the errors induced by vcrbnl dcscr ipl ion arc nol a consc-quence o{ int lospccl ioD pcf sc. Whi le this is c lear ly ln issue l i r furthcr rcscarch.severr lsources ofcvidcncc suggcst tb t verbaldescr ipt ion olnon verbal rcrno. ies do indccd iuducc rcf lect ivc processes. includiDg ( 1) rccogDil ion l imc folbw-ing verbLrl descr ipt ion is incrcased (Brown & Lloyd-Jonos.2{)03), (2) laccrccognit ion perfornlancc is both lcss uccurule Lrnd bngcl whcn individuals applyrcf lccl ive st f t r legics ( l )unr ing & Stcn. 1994). Lrnd (3) tho disrupt ivc cl lccls ol 'pr i l ) r vcftal dcscr ipt ion rc a( lcnul l tcd i l individuLrls rre l i ) fced to rnaks quickrccogrir ioDdccisions(Schoolcr&FlnSsl lcr-Sclnx) ler. I99{)) . Thus. the evidoDccis highly suttSest ive in indicr l inS thrt anr ly l ic int | . ()spocl ivo processos ir ld!co( lby dc\cf ibing nrcnl(r ' ios c. ln sorncl inrcs disrupt hol ist ic non verb l rccognit ioDpr(rccsscs- l { so. s lch evidcDco iD(l icr lcs lhr l thc vcrhuldoscr ipl i (nr procc\sos uroroL dr) iD8 dcqut l tc. iurt ico to thc nrcrx)r ics to which thcy afc dir( jc lcd.

Mind-waDderi lg

Inlrospcct ioDs brrr l r r r ind wrndof ing posc l l uDiquc chr l lc0Sc bccNusenrind r , ,andef iDg i \ dcl incd pfccisely by i ts r / rr1r1. ol r corrcspondcncc rvi thongoiug cx(crni l l events. Conscqucnt ly, thc now substrtr t i l body ol rcscnrchdcvoted to cxrmining mind-w ndcrirg has rol ied largely on iDdividu ls introspecl ive repof ls to r$scss wheD i t is occurf i rS ( fbf fecent reviews see (; iarnbrr,lg95; Scho(tef z/ rr l . , in prcss; Smil l lwood ?r r / / . , 2003). I low much conf idenceshould wc hxve in these rcports? In facf. when we cxlrninc lhc corrcspondcnce ofrcpor(s Lrf nind wxndering wit l l cxtcrnal evenls l ikcly to pronpl i ( . bchtvionrsl ikely to be induccd by i t . and physiological rcsponses l ikely to be associalcdwi(h i t , we l iDd good rciNons t() bel ievc prr( ic ipants rvhen they repoft thl t l lhcir

inds hrve wrDdered off. Ncvefiheless. recenl rcscarch also su€igesl\ thal p |tic-ipanls c n rcSular ly lai l to nol icc tha( thcir minds have wandered. Such fai lurcsag.r in i l luslralc thc importance lbr in lrosfcct ive analyses ol dis l inguishingbetween having rn cxpef ieDce !e$us kuowing th t oDe is ht! ing i t .

ll|idmLL lu the act:uru t- ol inttusPectiw rcPort: oJ nintl tL'tuuleringAlthough rnind wanderin!! episodcs arc by definili(nr urrelalcd 1(r thc prinary(ask, i l is neverthelcss possihie 1{) val idalc such fefofts by esl tbl ishing r

26 J.W. SCHOOLER & C.A. SCHREIBER

relationship between the nature of the primary task and the likelihood ofmind-wandering reports. Not surprisingly, some tasks are more likely to promptmind-wandering than others. For examplc, mind-wandering is more likely whenindividuals are involved in non-engaging tasks such as reading easy material(Grodsky & Giamb.a, 1990-199 | ) or working on well practiced tasks (Teasedale?f dl., 1995). Mind-wandedng also fluctuates as a function of sttess (e.9., Antrobuset al,, l9(t1), the gender of the experimenter (Singer, 1988), and citcadian rhythm(Giambra et dl., 1988-1989). Additional evidence of the validity of introspectionsof mind-wandering comes from the finding that rePorted mind-wandering epi_sodes are as$ociated with decrements in performance on a variety of demandingtasks including generating random sequences (Teasdale et dl, 1995), memoryretrieval (Smallwood er al.,2003) and rcading comprehension (Schooler st41, inpress).lntmspective reports of mind-wandering are also substantiated by evidenceofa correspondence between self-repons ol mind-wsndering and neurocognitivechanges associated with EEG (Cunningham et al., 2000), galvaoic skin response,and heartbeat (Smsllwood et al.,2cn4),

Evidence that introspections may overlook mind'wandering episodesThe consistent environmental, behavioural, and physiological conelates ofmind_wandering support the validity of individuals' reports when they indicate thattheir minds wandered off. But how confident can we be that, when they do notmake such reports, their minds necessarily remained on task? Bpisodes ofmind-wandering can also be rematkably elusive, as evidcnced by the frequencywith which virtually everyone reports episodes of mind-wandering during read-ing despite thc best of intentions and the fact that their eyes continued to moveacross the page. The striking thing aboul catching one$elf'zoning-out' duringreading is that had one realized one's mind had wandcred one would presumablyhave either stopped reading or stopped daydreaming. Thc fact that two mutuollyincompatible activities continue, strongly suggests a ldck ofmcta_awareness thalone's mind has wandered.

Schooler €, al. (in press) recently reported a mind-wandering during readingparadigm that was explicitly designed to illustrate the occurrence of initiallyunnoticed mind-wandering episodes (which they termed 'zoning-out'). In thisparadigm, participants read dull text on a computer screen and indicated everytimethey caught themsel ves zonin8-out. They rvere then asked various questionsincluding whether they had be€n aware that they had been zoning-out prior toreporting it. [n a sgcond condition, panicipants were additionally probed inter-mittently and asked to indicate ifthey had been zoning_out at that moment. Theresults revealed that participants: (l) frequently caught themselves zoning outduring reading, (2) were still often caught zoning out by the probes, and (3) frcquently reported that they had becn unaware that th€y had been zoning out' particularly when they were caught by the probes. These findings demonstrate thatindividuals frequently lack meta-consciousness of the fact that their minds hadwandered. even whenthey arein a study whcre they are specifically instructedtobe vigilantly attending to mind-wandering.

TII I ] PARADOX OF INTROSPEC]TION t7

Sumnery and ur?atsThc cvidencc lor the rcnl i ty of in lrospect ivc reporls of mind_waodering are supporlcd by dlc convergcnce of cvidencc lhat mind wandcring flLrctuatcs in pfcdictrble ways with thc cnvi |()nmcn(. is prcdict ivc oI behaviour, and corrcsp;rdstounderly ingphysiologicalresponses.Ncve.thetcss,duetothecmpi ical l t inacccssible nature of inner expcricnce wc tnust again maintain som; humil i ty ininterpret ing such data. for exarnple, i l is possible that in{ l iv idu.r ls. rcporrs ofdaydreaming episodcs are rcrually retfospective rcconstruc(ions. lndeed such anacco nt has been offered lbr drerurs where ii has becn suggcs(e{i thal Lifearnsta(es are nol thenrsclves conscious, but rnay sirnply be rc(rospeclively pcrceivcdas conscious st i r tes r l rhe i i r le of awakening (Dennct l . lggl) . Whi le suchaccounts cirnnot bc cnl i rely ruled ol l t , we rgain would appea] to the rcidcr,s f i rs(peNon perspecl ives. c ivcn your own expcricnces with calching your |nindwandcring. how plausiblc does i t sccrn that your expcriencc ol . thc daydfc: l tncoukl h ve bccr) ar i l lusior i . conslructed only af lcr you caught i l ,? When rtrob€haviourrr l . ervircnmenlal . and physiotogical evjdcncc Ior daydreaming expcnenccs is considcrod k)gcthcr with wh l wc cxpect to bo lbc strone f i rs l personrntrcspect ions ol rhe rca( lcr. rhc clse fbf lhc exFric l lce oI nind_w ndering isqui lc sl f t )ng.

Al(houlth thc cvidencc slr()n8ly sUggesis thal the occunencc ol-m ix l_wrndcrinj lcxpef lenccs ck)scly correspon(ls t( ] indivi( lut | l r 'scl l - reforts. rhc cvidencc lsr)fcvoit l r ia i luros i r) inrrospect ivc i rwtrrencss. l rdecd, the f inding that i r )( l iv j ( ju:r l \l fcquonl ly l . i l lo rcfort rnind wan(ler ing cpiso( lcs u nl i l thcy aro pr()bc( l \ l |1nrglysrrSScsls lh t mi ( l -witndcr ing opis()dcs l l l ly colnDronly l r i l lo ontcr ret i ll lw rcncss. l lcrc loo. howcvcr. we rnLrst bc cognizl ln l ofolhor pos\ ible lccoLrrr t \ol lh is l inding. stcrnrDin! l iorD ouf incvi tably indirect access t() othcfs ( l i fcctcxpcricDcc. I l is possiblc, l i ) r ex nple, thl l i r is no h l indivi( lurts tr i i t to not icclh l theirn) iddsh vewl l defedhurf t r (hef(har lheytbrgorth.r t thego.r lo l . rc dinSrs cornpfchcnsion. SLrch aD lccouDt crnnor bc ruled oul eDtirc ly nd is rgulhlyrnore phusible thrr lhc pr)ssihi l i ry rhrt | l ind-wxndc ngof j isodescouklbo fctro_spect ivc recrntstrucr ions. Ncvc.rhelcss. rhis may bc aDothcr placc where j l r )Lrybe uscl i r l 10 consul l the rcrdcr 's l i rst pcrson perspecrivc_ Lnrrginc crrching yourmiDd wxndciDg whi lc you afe rcading an imporlant docuDtenr. How fhusible i \i t that you could havc cxpl ic i t ly known that yr lur nr incl had wandefed off , bur torgotten lh l the goalofrelding is ro undcrsrand ? Wc cxpect that many rcadef\ rv i l l\harc our intui t ion lhat x more inrrospccrivelv compel l ing rccoun( is rhrt reaclcrsinlermit tent ly fai l k) takc stock of where their rninds arc curfenl ly residing.

Ilcdonic App|.aisal

Hcdonic apprr isal \ ( i .e. . assessment of one's own ple$urc or pr in) rcpresent yclanother domaiD in which rcse rchcrs are neccssari ly rc l ianr on iDdivi( lu i l ls.iu l f t rspect ions. As Mycrs otrserved, t lydef i0i t ion,thef inal judgcoIsomcot)c.ssuhicct ive wel l bcing is whoncvcr l ives inside rhat fcrson s skin. (Mycrs, 2(X){) ,p.57). Indeed. as iD the o(hcr domains thal wc hrve fevicwed. individ r ts.

28 J.W. SCHOOLER & C.A, SCHREIBER

introsp€ctive reports regarding their hedonic states are also often closely cali-brated with external cvents, related behaviours, and physiological responses.However, there are also times in which hedonic reports become urcoupled withsuch indices, suggesting that although individuals' feelings of happiness may beunassailable, their meta-conscious chatacterizations of these feelings cansometimes be problematic.

Evidence .for the accuracJ of introspective hedonic rcportsOne critical source of validation of the authenticity of hedonic introspectionscomes from the often close correspondence between fluctuations in envifonmen-tal factors and hedonic reports. For example, online mtings of noise-inducedannoyance (Schreiber and Kahneman,2000) and pressure-induced pain (Ariely,1998) have both been found to be close approximations ofthe current physicalintensity of the stimulus. With respect to positive experiences, Wilson andSchooler ( l99l) found that untrained subjects' ratings of how much they likeddifferent strawbeny jams were closely aligned with the overall jam quality asassessed by experts. Admittedly, there are some situations in which hedonicreports systematically diverge from physical fluctuations. However, in many ofthese cases the divergence between psttern$ ofphysical stimulation snd hedonioexperience are systematic and make perfect sense. Forexample, Kahneman ?t dl.(1993) observed that temperature-induced discomfon (i.e., cold-pressor pain)escalated over thecourse ofa minuteeven while the water temPetature was con-stant. However, this finding is not surprising when one recognizes that, whereasthe temperature ofthe cold water remained constant, the temperature ofthe sub-merged hand's tissue continued to drop.

ln addition to minorjng fluctuations in environmentalchanges, introspectivehedonic reports also regularly corrcspood to individuals' behaviours. Forexam_ple, there is a close correspondence between thc pain ratings that individualsreport during a cold-pressor task and the likelihood that they will remove theirarm from the water (e.g., Keogh & Herdenfeldt, 2002). Similarly, individuals'appraisals of the positiveness of objects is often, although not always (see dis-cussion below), highly corrclated with theit subsequent consumptive behaviour.Not surprisingly, people are mote likely to drink beverages (wilson etdl , 1984)and take home posters (Wilson et dl., 1993) that they rate more positively thonthosethey rate less positively. Hedonic appraisal s are also oftenclosely linked $a variety ofphysiological responses. For example, fluctuations in reported painhave been found to closely correspond to changes in systolic blood pressure(Hilgard, 1969) and to be well modeled by latent variables involving a combina-tion ofevoked potential, pupil dilation, nnd galvanic skin response (Donaldsoner z/., 2003). Physiological measures that fluctuate throughout the sPectrum ofhedonic experiences are somewhat fewer but nevertheless still exist For exam_ple, whereas the reported unpleasantness ofphot{)s is associated with increasedstartle reflex response, the reponcd positiveness of pictures is associated withreduced stanle reflex (Lang er dl., 1990).

THE PAR^DOX OF INTROSPECT]ON

Er idul cc./b r.lai lures ol ll.don ic i n I ro spe cti onsAlthough individuals ' h€donic rcpofts arc oftcn val idatcd by their convergencewilh othcr mcasurcs. thcrc arc sLlmc situalions in which these very sanre mer-surcs suggcst polclltial failurcs o[ introspection- For example, the correspon-dcncc bc[vccn rcporlcd pain and changss in lhe physicrl inteDsity of aversivesl imul i is markedly reduced when individuals are exposed to suggcst ions,inclucl ing hypnosis or phcebos, th l lead them to bel ieve that they should not bcexperienciDg discomlbr( (e.9., Mi l l ing & Dreen. 2(X)3). Simi lar ly, whcrcas Wil-son and Schoolef ( I991) observed a close conespondencc bctwccn individLrals 'repofted l ik ing of ja|ns and iam qual i ty, as determincd by cxpcrls, this correlr( ion was fcduccd io non-signif icancc whcn part ic ipants analyscd why they Gltthc way lhcy did ib{}Lr1 lhc. ialns.

In adLl i l iou (o or l inc nlcasurcs. r numbcrofsludics have l i ) nd that in lrospec-( ivc roports ol lhc rclrospccl i !c globr l hcdonic vrhrc ol episodes l l i l to f i r l lyref lec( whnl wi ls aclual ly exper ' ienceLl mornen( by rnonent (even when the epi-sode is br iel rnd the inlrospect ion occurs i rnmediately al ler the exp€r ience ends).I-ogical ly. the tot l p lensure (or discomli)r t ) der ived f()m nn cpisodc shouldrcf lccl lhc sLrm of thc hcdonic vr lucs of thc nnrnlcnts thrt compr isc lhc cxpcri-cncc, that is. thc inlcgrr l of plcasurc (or pain) ovcf t inc. Kahncman . l ( / / . ( 1997)l i ) rmir l i lcd lhis f f inciplc rs r nofmal ivc {hcory ofcxpcricnccd Uti l i ly. hul s lu( l ic\ol lhe rohl i r)nship bclwocn prt lcfrrs ol-rrx)mcnlary cxpcricncc aDd rclrospccl ivoovulur l ions o1-episodcs hI lvc consislcnt ly found that choiccs lDd introspcct ivcglohul . j rdgrnonls ( lo r)(n col l l i )nn ro {his tr inciplc. Spcci l icr l ly, this rcsoruclrl iD( ls lhir t individuir ls rc l f {)spcct ive ovr lurt ions ovcremphrsirc lhc plclsLrfc ( ' fd isco| l r l (n1 I l r thccl) i \ (x lc 's r ] r)st cxt lc[ ]c r )nen( anci r t i tsending. lhc pcrkcDd rule' . Othcf rnomcIts. and oi lcn lhc cpisodc's durat ion. havc l i l l lc c i lccl onglob l hodonic l l \sessDrcnts. Al though thc pctk cl ld rulc rnay not cnt i rc lyexhi lLrst lhe inl lucnces ()n glotral hcdonic cvr i lut l ions, lhc issignrncnl ol grcl lcrweight to lho perk nd end rnornents has bee found for evaluat ions ol plcasantarrd unplcrsa t v idcos (Ffc( l f ick\on & Ki lhne|niLn, 1993), the discomfi)r t ofnrodic l pf t)ccdurcs (Rcdcl c icr & Krhrcnrrn, I996). lahotutory pr in indLrccd bypfessufe 10 the l iDgef (Ar iely, 1998), chronic arthr i t is pain cxpcricnccd ovcr thccoursc of scvcral drys (Slone ?t dl . , 2l) l l ' ) , and noise- induced rnnoylnce(Schrcibcr & Kahncman,2000; Ariely & LoeweDstein,2(X)0;Schfeiber.2(J{) l ) .When cornbined $i th subiccts bcinei l i t i lc inf luenced hy the durat ion of an expe-r icncc. th is pl t tem ol evalulr t ion by peak and end can lcad to inlr{ 'spect ivejudgmcnts and choiccs th t dcvixte f f (rn any r t ioDl l or objecl ive ccounl ol wh 1was experienccd lKNhrrernan ?/ r / / - . 1993).

A sccond soufce ofevidence l i r r lLr i lures ol hcclonic introspccl ions comcs frostudics rcvcal ioS discrepancies between hedonic appraisals rDd related behtviours. A vrr ic ly of ! rr iablcs hx\ heen f i )und 10 conlr ihute to snch dissocir l ions.One soufce ol such discrepincics is analysin8 the feason for hcdonic expericnccs. FLrr cxlmple, Wilv)n .r al . (1991) found that $4rcn individuals analyscdwhy they fc l l thc way ihcy did.rboul var ious ar l posle$. they were t I rcre l ikely (oselect posters thl t l they ul t i r tatcly did rot put up on lheir wal ls than they wefe to

:10 .I .W. SCIIOOLER & CA, SI]HREIBER

makc such choiccs wheD they did not analyse their feel ings. Discrcptrnciesbetwccn bohaviouf and inlrcspectioD .rhout affcc( can lso result ffom sublirninally prcscnted rf}-ectivc stimuli. For ex.unple, Strahan rr.rl. (2(X)2) found lh.rlindivjduals who wcrc subl imin l ly presentcd wi lh sad lxces were no morc l ikelythan conlrol subjects 1o repoft reduced mood. Nevertheless, thcy wcrc tnorelikely to cngaSe iD stratcgiss lhar xre ffsociatcd with mood repaif, such rs choosi Ip lo i i . ter r , ' ut , l i f i i t r r thrr l lun ct j t ) rnu\r . .

A third source of cvidence of tni lurcs of hcdonic in(()spccr ion comes l ionrinvest igal ions ol discrcpancies between hcdonic rcports and physiological mea-surcs. l l i lg rd ( 1969)observed that whcrcas hypnosis attenuatcd thcpain ssoci-atcd wi lh keeping oDc s nn in cold watcr. i t d id not at tenuatc changes in bloodprcssLrrc thr l rrc lypical ly highly corrclatcd with pr in in lhis task. l here are .r lsoccrtain dist incl populat ions ol individuals who consislcn( ly show dissociar iLrn\bo(ween physiological rcspoDscs i [ ld al ]ect ivc i rr l rospcct i rJns. Fof examplc,whcn shown slrc\sl l l v idcos i r( l iv idui l ls ident i f icd as'rcplcssors' r€port lcssstross than c(nr lrol subjccls. whorcrr their physiological rcsponser (c.8.. galvaDicrkiD rcsp()nsc) suggest hiShcr lcvols ol stress (Ascndorpt & Schercr, l9f i l ) . Sirn,i l r r ly, Adrrrns d ul . l l t )96) l in r l l thnt horrr{}phobcs who rvcre showD cxpl ic i lt turvics ol int l iv iduals c|glging in hornrrscxual rcts roporlcd rn abscncc ol sexuir l )usr l , whi lc physicr l ly evidoncing coosidcrablc i rrousal as rrrci lsurcd byponrle lunrcsccuco. As l unlcist i i r dr . r1. ( 1998) obscrvo, th is hl tcr l i rdi l lg raisesthc parrdoxic lqucst iolr ol hov docs sornc0l le DunLtSc Io lccl scxui l l ly turncdol l whc| l his or hof body is oxhibi t i r ) ! i . r s l forrg posir ive r f l )usr l ' t AgriD, rhe disl inct ion bctwecn expcricnt j r l consciousncss and Ineta-consoiousucss t l ty ol lar apotcnl i r l i |n\ \ 'vcr. Accordingly. iDdividLuls nt i ly cxpcricncc the af i )us l bur.bcc use ol thci f strong rnot iv l ion, l i r i l ro becornc Incl -aware ol thir l cxpcr icDcc.lndccd, s rvi l l t rc urgued in lhc l lnr l secl ion. thc disl i rcr ion betwecn corscious-ness and incl i r-oonsciousncss Driry iLccoL!t l lbr rhe vaf iou\ s irul l t iLrns in whichindividui l ls iutr(rspect iorrs disconnect t ionl lhc cnvinrrrntent l l , bchavioural , i rndphysiologicr l indiccs.

Sumnrrtr\i und .:uyeatsAs in thc olher domains reviewed, thcrc.tppears to bc compel l ing empir icalcvidcnce lbr both the succcsscs rnd lai lures of hcdonic iDtrospect ions. On theposit ive side. thc consistent nlanncr in which hedonic .rppraisals map olr lo envi-ronrnenlal f luctu.r t ions. bchavioural re\ponscs, aDd physiological nreasufeshelps (o rcassure us thr l rhese reports do have,r fbundat ion in experience. At thcsamc lr l ie, we must also rernain wary of the possibi l i {y of ahenrt ive accounrs_For cxarryrle, it sccrns plausiblc tha( ar leiNr somc of (hese co|reslondcnces rreno1 due to f lucl at ions in individualr ' cxpcr ieDces per sc. bul rrrher in thcirbel ief l \ xboLrl lhe expcricrrces. ID al l l ikcl ihood, iDdividuals hrve rheories aboutwhen hcdonic st res should be berteforworsc rhar would also beexpecred to corfespol ld 1() reportcd I luctuat iols iu hedonic valcncc (e.g.. individuals presum-ably believe tha( changes in the lempenrurc oI exlremely cokl water shouldcc'respond to changes in pr ir) . Neverthclcss. we again woulcl appeal to ths

THE PARADOX OF INTROSPECTION

reader's firsGperson perspective as one additional source of cvidence 1{] consider. Does it seem plausible th.rt, when individuals report cxcruciating painresulting fiom keeping their arm submerged in cold water for extcndcd pcriodsof time, this simply reflects their theory about what they think they are feeling?While it may not be possible to rule out such an account, given the vividness ofthe experience ofpain, we expect the reader may find such accounts implausible,at lcasi ln many cases,

Whilc certain crrors in hcdonic rcports secm untenable, we argue that otherssccm morc likcly. Cascs in which individuals' hcdonic reports systematicrllydiverse from all other measures raise serious doubts .rbout the accuracy ofthe introspections, and suggest dissocirt ions be{ween experience and meta-consciousness. Here tu) some caution in interpretrtion is necessary, as wc canrever be certain ifdiscrepancies between self-reports and otherpotential indicators ofcmotion rcflcct an unconscious emolion (Bcrridgc & Winkielman, 2003)or a failurc for an cmotional cxpericncc to cnter mcta-awarcncss, Neverlheless.whcn wc apply our own firsl-pcrson pcrspeclivc in considering, lbr example,rcprcssors' failurc lo acknowlcdgc any fcclings of arousal, we agree withLambic and Marccl (2002) in concluding lhat ' i t is implausible that, lbr someone|| ho is exhibit ing emotional body states aDd behaviour, that there is "nothing it isl ike" to be and behave in such u way' (p. 250).

Why Do lntrosp€ctions Sometimes l'ail?

[Jp 1o this point our primary Sonl h s heen to persu{de the rcadcr that by consid-c.ing how irrtrospcctions co-vuy with othcr indices it is possiblc to makc a rca-sonilblo approximation of whcll in(fospcctions arc trustworthy and whcn lhcyshould he viewed wilh some skepticism. It€r iodical ly, we have also sugllestedthat thc dist inotion between consciousDess rnd metFconsciousness nrry help ()conccptLrl l l ize these dissocirt ions. Horvever. we hrve lareiely delayed discus-sions of lhc mechanisms that nlry lcrd t() f Ir i lures of introspcctivc rcports. In this\ectioD we lurn l t} lhi\ issuc.

The basic idea underlyi[g the disLinction bctwcen consciousDess and me1 -consciousncss is simply that individurls otien have experiences without necessarily explicitly inrrospccfing about thcm. As a consequencc introspcction canfail lbr two very general reasons. First, introspcction may not be invoked. Suchintrospcctive failures are what Schooler (2{X)2b) relers 1o as 'temporal dissoc;a-tions' in which cxpcricncc occurs in the absence of meta-awareness, People'slailure 1() notice thal they are zoning out durinB reading is a good examplc ofrtcmporal dissociation.

Second, introspcction can fail because, in their atteDpt to characterize anexpelience. indivicluals may distort i t- The maiority of the introspective ditTicul-!ics descrihed in this review can be characlcrizcd as lranslation dissociations'bctwccn consciousness and nretir-consciousness. Some of the sources of theselranslation dissociations nray result fiom processes associated wilh .1cr?.Jior,t run I lb r nra t idr. 'JntJ s u b s t i tut io n.

3L

32 J.W. SCHOOLER & C.A. SCHREIBER

DetectionAlthough many experiences in life are salient, others are more subtle lfan expe-rience is very subtle then it may be difficult to notice its defining elements Putanother way, the 'siSnal', which introspection must detect and interpret, is weakor ambiguous. This may explain why individuals can fail to notice how theybuild their mental images (Kosslyn et dr,, 1998) or that an image is missing aminor detail (Chambers & Reisberg, 1992). Similarly, the several studies thalused subliminal prese0tation ofnegative stimulito induce dissociations belweenhedonic introspections and behaviour may have been successfirl beoause theirmanipulations werc so subtle that participants failed to exPlicitly notice theaffective change (e.g., S ljahan et al.,2OO2\' lndeed, when a hedonic experienceis sufficiently subtle, attempts to 'look' for the positive elements ofthe experi_ence may actually interfere with our ability to extract thoseelements (Schooler€ta/..2003).

TransfomarionsAlthough some introspective tasks appear to be accomplished with relatively liFtle di$tortion in the presence of an adequate signal, others rcquire interveningoperations fbr which the system is ill-equipped Both the analysis of perceptualwholes into component features (as when subjects provide detailed descriPlionsofa face or a flavour) and the synthesis of a set of moments ofexperience into anepisode (as required fot Slobal hedonic evaluations) are examPles.

One important soutce oftranslation dissociations stems fiom the inherent non-verbalizability of many experiences. lf an experience is holistic ond ineffable,then attempls at decompo:'ing and trunslalin8 that experiencc into words m3ydiston it in systcmatic ways. This may be at least part ofthe reason why desctib'ing visual images can irnpair later performance that relies on those images (e g.,Schooler & Engstler-schooler, 1990).It may also explain why, when individualsatlempt to analyse verbalty the basis of affective experiences that arc inherentlynon-verbal (e.g., the reason for liking a painting), they can genelate appraisalsthat fail to correspond either with the physical qualities of the stimulus (Wilson& Schooler, 1991) or with participants' subsequent behaviours towards the stim-ulus (Wilson et al,, 1993\. In all of these cases, individuals are required todecompose non-verbal holistic experiences into words. This decomposition pro-cess may both lead to the generation of specific inaPpropriate content (e.9.'Schooler & Engstler-schooler, 1990; Meissner €t al, 2001) and/or cause indi-viduals to adopt featural/analytical processing styles that are inadequate fordeal-ing with holistic non-verbal experiences (Fallshore & Schooler, 1995; Schooleret dl.. 1997: Schooler. 2002c).

Just as antlysis of experiential wholes can introduce systematic distortionsinto introspective rcports, so too can synthesis ofmoments thatcomprise experi-ences that exlend over titne. when individuals attempt to take stock of experi-ences that occurred in the past, they are, of course, vulncrable to the variouscognitive distortions associated with recollection. Whed a judgment requires

IHI ' - PARADOX OF INTROSPECTION l3

aggrcgating many moments of cxlcricnce. however, the potentiltl lbr distortionincrcases. For exumple, al though memory I imitat ioDs m y introduce error in(oretfospect ive judgments of ihc intensity of r pa i !ular momen( of noisc cxposurc. forming thisjudgrnent is, ncvcrthclcss, a 1af less complex construction thanjudging the 'total amounl of noisc' to wh;ch one was exposed over soneextcndcd t ime period. ln the casc of hcdonic judgments, due to an iDabi l i ty toconpule thc intcgral ofmoments ofpain/plcasurc ovcrt imc, people fbcus on themos( sal icnt aspccts ofthe expef ience (pr imari ly, the peak and the end) aDd givel i t ( le wcight io thc renr.r ining elements (c.g., sonret imes even neglect ing theexperience's dural ion).

Inlr{}spcc( ion may sometimes go wry bccausc thc intbnnat ion accessed is notthe recnld of i rctual expef ience. Instct ld of rc lr icving the memory ol consciousexperiencc, pcople may instend br ing bcl icfs to nreta-awarcness withou( .e l izing th ( (he contcDts of rnct i r-consciousness may divcrgc suhstant ial ly f r(nn whatw s experienccd in consci{)usness. Expecl l rncics and mol ivat ions:rrc amons theibrccs iDvolvod iD lhis lypc of el l . ( rr .

I r i rsr. whcn individuals hrvc stronS cxpectrt ions i rboul conscious cxpcricnccrhcy nray l lcccss the expcc(at ion rr thcr thrn {he i lc luul oxporicncc. Such irnirccrnrnt nrry. r t le ls( part i r l ly. cxphin the impLrct ol hyProsi \ and plrrcchos(l t i lga(1. 1978). Accordinl t ly, i l individuals Seorinely bel ieve lhir l (hoy shoul( lrr(n bc lccl ing pr in, thcr l thcy rnay bc lcss l ikcly l r) nol ice i l . A si ln i l r f lypc ol suh-sl i lLrt i r )n ml ly undcdic thc l r0nslNl i r)n crfors nssoci i r lod wi lh mol ivrt t ion. lJDdcrccrt i r in circurnsl i rnccs. individur ls In. ty be motiv lcd lo i8norc lhcir subiccl ivol tr l | rc, rhcrcby. i rvoidinS hrving tr) cxpl ic i t ly dcI l l wi lh i ls inrPl icat i (n\ . Anunwil l ingncss to acknowlcdgc thcir cxpof icncl j to lhcrnselvcs mtty cxPl l l in whyho ophobcs lr i l to nol ice the rousal thr l thsy ( lcf ive f f tnn vi€\ ' iDg horix)scxuri li rc ls rnd why rcprcssor\ i r i l 10 rcpoi thc strc\ \ they expef ierrce whi lc vicwinS(l istLrfhing rnovics. In crch ol lh l jsc cascs. individlral \ ' nn) l ivat i l )n to avoidaclnorvlcdging unacceplablc cxlcr iences Ina) cirusc thcm to iSnorc those expe-r ieDces and rcplace them wi lh nlorc i rccept ble tccounls (scc also Schoolef.2{)01). The disl inct ion hetween conscioLlsness and mela-consciousncss thusol l i$ rn al temative thrcshold lbr del ining l l tc avoidmcc t t t r ihLtted () rcpres'\ ion. Accordingly, f t r lher than banning thoughts and lbcl ings fo thc unconsci()us.nr(nivat ion processes may prevcnt individuals l iom not ing thc occurrence ol 'rhoLrghts ilnd feelings thilt actually do comc to mind. The coDlents oi mctacoDsciousncss, in ruch cascs. mny correspond to prcfcrrcd helief-.s mlher thanrctualexpcncncc.

lmplications lbr Mef hodology

In xddi l ior to i ts theorel ical valuc, thc ident i f ic l ion ofspcci f ic condit ions underwhich translet ion dissocirr t ions arc pa(iculady tp( 11) occur also off t rs method-ological impl ic l t ions regarding techDiques for maximiTing introspect ivc accufxcv. The trrnslat ion dissoci: l t io lrs iNsociated wi lh norvcrbal iTable experier lccs

J.W SCHOOLER & C. . SCHREIBT,R

sultgest that fcscarchcrs eithcr {a) lbcus on morc verbalizable experienccs (e.g..logicr l problem solving. Er icsson & Simon. | 980; 1984; 1993.), (b) rrain paf l i ; i -pants k) dcscribc rheir cxperienccs rrorc efl-ecrively (Jack & Shallice, 2001;Melcher & Schooler, in press), (c) rcach people mindlulness practiccs (e.g.Brown & Ryan,2001), to enablc them to become morc consiste;tly attuned t_othcir expcriences, (d) force quick rcsponses that minimize vcrbal rcflcction(Schoolcr & Engstlef-Schootcr, 1990), or (e) offer pcrceprually g()undcd non-vcrbal fcsponsc optiors that adequately capture the breadth of the experiencc(c.9., Barroshuk, 2000). Thc rr nslat ion disrociat ions associatccl withjntrospecling about subtle expcriences suSgesl that rcsc.rchers should be espccially wary of such inrrospections and/or find ways to increase thc salicnce ofihccr i t ic l d imensions. Thc lrrrs lar ioD dissociat ions associuted wi lh expcctrncicssuggest lhat fcsearchcrs should r t tcnpt to idcnt i fy individuals. expcctancies anLlassess the dcgre€ ro which f luctral ions in expectancy mcdiate i l l rospecl ivereports. 'The problcnls with jntrospcct ions based on tetnporal synrhcsis P() in( (othe nccd l i ) f nlethods to assisl people with more ful ly i r rcorporat ing tel ; icvableint i )rnrat ion i r(o thcir global hedonic evalual ions. Final ly, r l to translat ion disso,cr i r t ()ns rcsult in l t f iun nrot iv l ion should bc evaluatcd by dctertniniug whclherindividuals possess I luol iv i t t ion to ignorc or Inisreprescr i t their cxpcr ioncc, andl l so, rese rchers Dtusl cxalninc whethcr di l lcrc[ocs in this rnoi ivut ior al tcrsel f--rcports. In i r l l c i lses. 0s i l lustr i t ted io the prcceding soct ions, conl. idcnce irrintrospect ions can be gained whcn thcy re l i )und 1() systcmatical ly oo,vi l ry wi lhexpcnencc, beha! iours, and/0r physiologior l rcsponscs. On lhc olhcr hand,conccnr is w rr tnled whcr thcsc indepcnden( indices bcgin to divergc l iom thcintrospccrions ro which thcy should olherwrsc cofrcsponU.. Of course. lhc l ic l rhi l ( wc can conccptual ize var ioLjs l l r i lur€s of i r j t rospccr i(nrin tenns ol dissocidioDs bclween consciousness anu mcl i l_consclousncss doesDot rnean thi l r this app|oirch musr Decessari ly cxplain al l such fai lures. In srrrnccascs faj lurc-s of i l rospccr ion nl ly s imply rel lccr part ic iprutts. unwi l l i tgness lobe tbr lhcol l l iDg. For exrr]rple. rcpressors rnight know thcy arc :rrouscd iuL sinr-ply.be unwil l jng lo adul i l i t . For luni l tc ly, thcfc are sorne techniques for explor ingsuch possibi l i r ics. For example, individuals have sorne (albci t inpcrfcct) iortrr i lover thcir f iLcial expressions (Ekm.rn, l99l) . I f rcpressors di( l know thcy werc{rcusecl, fhcy would bc expcc{cd to try () cont()l their lacial erfrcssirnrs olarousul, sornething thcy havc not becn obscrved !o do (Ascndofpi & Schcrer,1983). Moreovcr, therc are var ious scales thal c n hclp to idenri ly that subscr oiindividur ls who may bc moderal ing their responses for sel f :presenrat ionalreusons (e.9., Plarr & Dcvine, 19913).

Evcn i f individuals arc nor dcl iberatcly mislcading rcsearchcrs, rhcir fui1urc roaccuriLtely rntrospect about psychoJogical crenrs does not nccessari ly imply rhatthose evcnts wcre expcrienccd in thc absence 1)f mcrr_ awi eness. AD;thc;;ossibility is th.rt thc psychological evcnr nevcr crossed the bouDdary {Jf coos;iousness. for exan c. Iterfidge & Winkielrnan (2003) argue rhar lhe occurrencc ()1.behavioufal el ]ects rJf ctrot iooal manipulat ious (e.g., changcs in cousumprivebehrviour fol lo$, iD8 subt iminal ly prcsenred srd faces) in ihe abscnce oi any

THI] PARADOX OI- INTROSPI'CTION t5

self rcported changcs in eDr(nion implicate nonconscjous' (i.e., not expcrienccd) cmotions. Whilc the notion that cmotions can go entirely .unlelt' is (fromour rcspective first-pcrson penpectives) somewhat dillicult to conceive, it cannol be conceptually ruled out. Tt is cxtremely difficult 1(r distinguish betwecn aDinternal event that is expe.ienced, bul not explicitly noticed, as opposed to onethat is not expericnced a1 al l . At a minimum, the dist inct ion bctween conscious-ness and nrcta conscklusncss provides an altcrnative to the standard vicw thatnon-reponedpsychologicalcvcntsmustbenon-conscious.Ult imately,however,discriminating belween whethcr something was experienccd but not mcta con-scrous. vcrsLls not expe. icoccd a( al l , wi l l rcquire the idcnt i f icat i r of metrsufesthat can i l luminate the cxpcricnt i t l qual i ty of an event. In this regard, i t is worthbrief ly considering how Sperl ing's ( I960) parr inl reporr parrdign cnhanced oufknowledge of thc cxperience of bricfly presentcd images. There was a timewhen, despitc inlrospecl ions to thc contrary, thc cxperimental dxtd suggcslcdtha( individuals wcre only conscious ol fbur to f ivc i tenrs in a br ief ly f lashcdrrray, f f this is al l people wcrc able to report (c.g., Morton. l94l) . Howcvcr,using the part ial report parndigm ( in which part ic ipants wcrc cucd to repo|t . iusl aport ion of a previously pfcscntcd visur l . rrray) Sperl iDg found thxt peoplc aclur l ly cxpcricnced the lul l arruy, ls evidcnccd by the i rct lhat they wcrc able torcport al l o l the cucd sccl ion even thouSh lhcy did nor kDow :t1thc t imc thLtt thcarray w. ls vis iblc, which po(ion of i r thcy wcre l io in8 to have to rcporl .

The pft)ccdurc uscd by Schoolcr 4r a/ . ( in prcss) to explore discrepuncicshctween sel l 'caugbL nd pft)be,caught cpisodes of zo ing{)ut dur ing ro dinSrniry also hold pfol l l ise l i ) r rcvc:t l i l lg experienocs associatcd with olhorwiscnon'rcportcd psychologic: t l cvcnts. As with Ihc pnrl i l repof l pnradigm. pcr iodicprobiDg oi expericncc cnablcd p r l ic iptnts ro not ic l j br icf ly cxperienccd cvcnts(rniDd-w nderinS cpisodcs) bel i )re thcy wcrc l i ) rgottcn. And l ike thc cornplr i -sons ol fhc f i l l vs. part ial rcporr proccdurcs, comparisons of scl f-caught vs.pr)be crughl cogDil ive eveols rcvcul the degree to which expcricnccs can bcovorlookcd i l thcy re Dot promptly rcpo(ed. In pr inciple this proccdure couldbc rppl ied 10 revc. l l othcr rcrnporal dissociar ions of mcla rwareness includi l lghi lures 1o not icc unwanled thoueihts (Wcgncr, 1994) or cmoti{mal cxpcdcnces(Lane, 2000) where expericnccs may f l i l to cross the threshold of Dreta,

Al lhough () dalc, no equiv lent proccdure has been dcveloped t() vr l id.r telr lLnsl i | l ion dissocial ioDs. ul l i lnately i t sccms l ikely dlar conceptual ly rc lalednlanipulal ions nl ight be dcvclopcd. On the assumptioD thrt individuals cannotrcport episodic mcmorics for states that wcrc nol experienccd at some levcl. ilmay he possiblc to find retrospective cvidence f()r situations iD which mclaawarcness ini t ia l ly rnisreprcscnrs experience. For cxanple, i f incl iv iduals werealerled to the tact that lhcy rnighr hlve been sublininally cxposed 1() eithcrneg.rlive or positivc faccs crn a pd(n occasion, and/oriflhey reccivcd some rypc ofmindfulncss training (e-9. Bfo$ n & Ryan,2003), rhey mighl reveal a reirospcc(ivescnsiti!ity to changes in thcir affcctive state that thcy did nor notice at thc rinre.BvideDce that individuals wcre caplble of retrospcctively acccssing orherwise

.1.w. SCIIO(n.ER & C.A SCIIRLTBER

unrcported sl i les would pfovide onc potent ial way of docutnent ing (hl tnon rcported psycho logical eve nts were. nsvc(heless. cxperienccd at sdne lcvel.

Givcn the di f f icul ty oI dist inguishing bqween psychological cvcnrs thar f t i lto rcach meta awarencss from those that ncver makc il iDto consciousncss, wemight rcasonably wondcr how inlport t ihis question really is. llowevcr, hereagrin is a situalion whcrc the paradox of introspection bccomes critic.ll_ F.om lhethird person perspective of empiricrl scicnce, what mattefs is whitt is firelsur_?rblc. and to ihe degrcc lhat expcrience itscll defics lnelsufentcnt. its rclcvancemrght seem moot_ Howcver, f rom I imt-pcr lon perspcct ive expcrience is cvery_thjng. By analogy. froln the pcrspectivc of the succcss of lr surgsry, it lnay makcl i t t lc di l fefence whcthcr the pat jcnl was rct l l ly unconscious or simply ul iable k)reporl / remembcr thc cxpef icncc. Howcvcr, j rnn lhc perspccl ive of thc pnl ieDi.this di t lercDce matlcrs considc.ablyt l hus dcspiro rhe inhcrsnt di f f jculry in dist inguishing betwccn psychological cvcn(s rhar fai l 10 cfoss t l le rhrcshold ofconsciousness vcfsus ntcta-consciousncss. advrnccs oD this issue wi l l bc cr i t icul forrcveal inS thc naturc ol cxfcr iencc both 1{) scicnce irnd ro ourselvcs.

A c kn ( ) v, I k I !:. t, k, n !.\' fhc wri t in8 ol this rr l ic le wl ls sLrpponcd by a Srt l rrr ro rhc l i rsr i rurhor t . rorn thcOIi icc ol t tdrrcl t i (m, i rnd hy an Individur l N l ion l Resc rch Servicc AwIIdBrt lnlcd to thc socon(l aulhor hy the Nlt ionr l Insr i lu lo ol Mcrrrr l l {cal lh.

l{elcrenrcs drrr \ , l l l l . $r lAht. l , .W & t . .hr . BA 1l9rr6). k hrr j r ) rh$ir n\v{ . txrcd s i th t roDrrs.r t r r l

1. . \ ta.r . t , t . t , , . , t ; . l r i |1. , r , r\ ' .1.1, I l? ,1,- . Ar"" ' , r r . . , . , . / . . , . . , . . . . , , .ld,i/l1.115. fli :,1(r l7

Air . l r . l ) ( l I r { )8) . ( inrbi i i ' l ! er | t | . feeir !cr . r tnr , theci i . t r ,n duftr i ,D. intcnstry c l rnrrc\ r ll i i i i : i l j " l l ' " ' - '* " ' \ '"rrv'r ' i 'cv.Lr!x' ' trr\ ' ,J ' , , , ' t , , t u,hu\nnnt)u^x, 'Mtkit ' .

Ar ie l ! . D & l .oewcr( . ! r . ( i 11000). Whcr ( id.s. l rhr I jn rn.r lcf in rulrrnefr dni t ( tecis i (nr dukm!t . ..tul,rr1.lt:\rrrii"rrLn l)\tjr,r,(| a;,,.nrl. t29, pp 5l)8 lt.\ ' , , , . .1. \ ' . r ' . r , t ( A\rrr . , t t , r . , \ , , . , t , . r . , r , f .1 ,"raf . . t . . - \ t . . . . . r 1,r ,1

f l r l . l . \ nr ' f ' , rJJ, t . , . i r - r r , , t1, , r1t . , ,11. .n. / r . r . , / , i . . J t t1, .d\- , .A.n. l . r t l . r !

^ \ .1 '1, .11< "qt ' . t t . ,1, . . . .n. . , , , .1r . . , , .D . . r . , | . r . , \ rvr(r i , , { , , , \ .1\1r; . '1\kr j J ' l r p ' i . ' . . . ' i . . \ . , , . , t t . i r . l , . . .b. t , r . . / . , r r , l

/ i "1 J , lL , , r ' / . . , . r , rJd 45 t . |a t , , r r ,rldrlosl k. L M. t2oaJar). I\ych.0hr-si.rt rdvrnces x rrrc slud! ofgeneri. vrrjdi.n D lxnc.. tnzft,.l:1. p ll)5l i rmeitr . ' , l t F. t ) l c , K. & S,rnmcr. } i . t . ( l r t )8) . .Fr.rdnn. ic i i rNc me.h!Divn\ nml cnrt i r icr l l . tndrtrs\ rn m..rrrn !)ciil pstchol.g!: Rer.t .n ldmrlt.i. Froic.riinr. {lj\ph..njcnt. undoiDi. i(rhln)n.'.

n'rtiuldr)n xnd den irl , /.,u\tul .l t'rrtuuh r,66, pt \)at \21Berridgc, KC. & Wmkr lDijr, p (2ll0.l ). Whxt ii I n uncoms.t.tr \ enoriM I ft.he cxr. n]r unc.ns. iqi \' ' \ l t ie ) . ( ormhh anl I :ht r r , , . l?, Do | j t 2t lBrndrnonr.. M A.. S.lunc..l W. & Cix66tnd, p. {lgglt. A enuxrtn! rcrbrl ovcrhidowina lhroulhvl$al (1rr\lr .e\ ..r,,ut ol Lrytirunlt I,t,\hoh[\ ttuni ( M!,ary tt t (..!;nn,Li3.

pp 915 . l L|n.xr, C & l-byLl r.ies, T.t {20011, .Vc,brl ovcrhato$nr! .i dtrnliplc id.c rtrd car r..rjrnrlion:EIJc.r\ or lnhii \ersn r.ross tdtsorl vcrbxr d.a.flpridi\.. A rlr,l/ t:t,prdr p\.tt.t;!;. i,l l , .qn.KW &lt txn.R.M {201): l ) . .Thc hcnci i lsothrtn! pEs.nt : Mjn(] l t r tncs\ ! id r ls rol . i i f \ycho

, l , , ' l { - l , ' t , \ , . t , - - t ' , . . . ,4. t J . . , . tJ t . , , r , , . .{ t , . ' . ' t l r . / tod, . / r , . r r . . , , . , , r , . , r , , , , . / / . . , , \ . r \ . r o.r , , l

31

Chamb.(. D. & Reisberg. D ( 1992). What an image depicls depcnds on whar an inagc m$ns. , Co,l"itjrP P\t"< holob", 24, pp. 115-11.

Cunnineham.S.. S.erbo,M.W. &F.eenan. F.C. (2n00), The eleclrcconical corclates af daydrer ningdudnA vigildnce laski, J,rrMl if M.btul IntKry.U,pp.61-72.

Dennerr, D.C. (1981). Aie dreans expericnccs? , in D.C. Denn (, Braintrorns: phibsafii.al Elsorson Mind and Psr.:hak'st (Catnbridge, MA: MIT Pre$).

Denner, D.C. (1991), Ca,& r,6&$ Elrkt.d(Boston, MAil-ittle, Brown and Co,l.Dcscanes. R. (i617). Dis.,!/s. an lll. tnl.l'^tis.Donaldson, G.W., Chapna., C.R., Nnlemura, Y., Bradshaw, D.ll., Jacohson. R.C. & ChaDman_ C.N.

r2uorr . PJin and rhe dcten.e rc\p, ,n.e: Srrucrurct equ. ' t rJn modcl ing re\ert , a r .n,dinlrdpsy.hophysiolo.,:ical response lo inc.casing paintul stinularion . Pai,. 102, pp.9?,108,

Dunnrnr. D. & \red, L.B r loo, ' . DFUnrJnhinp J(curr 'e ' r "m

mrrumFe}e}rrre. . idcnr. f icJrh,n.via inquiri* dbour decision proccsses. Jotnal of Peno&litl and Sociat ps!.holo|r,65.

Ekman, P. (1991).7c1ling ri.r: Clr.s to D.c.it 1n th. Md*etpla.c, Politi.r, atul MattiaR. (Ncw f.tk:W.W. Notun and Co.l.

Fr i t \con, K.A. {2rn2 ' . Inwrrdr o focedure lbrel ' r r t inE rcr l -a lcrpR, iun. ' l non-rcrb. ' te\peIcr. (e$i lhourre( l .vry Inrerprer int rhc !crh, lo\erhJ, l ! ,*rnE el l<r q r l in lhe lhc(treU-Jt t rJrn($or^ turptutocol anrtysis', ,4ppli.d Cot h itir. Prclnlog\), 16. pO, 91ll 'l ,

Eric$on, K.A & Sinon. I LA ( 1980). Vcrbal rcpoft as drta' , Psychtb|i aL Ratks,tl7 , pt,.2t a-5t ,Eric$on, K A. & SinDn, H,A. (1993), P roan Analtti!: Vr,fl,/l n./,rls,4r D/rd (Cnmbrid8c. M :

Fallshoe, M. & Schoolcr, J.w. (r995), vcrhal vulnerabili(y ot perc.pt@l axpenise , Journut .t t:rp(inl.fltal ,\y(:tunqy: L ning, M.n.ry, and Cosnition,2l, pp. ltotl 23.

F dh. M.J. (1995r), Cutrent issues in ncnmpsycholosy ol imagc gcnc|llion ,,{a!ralsy.r,r,/jnr, t3.

F&rh. M J. (1995b). r'hc ncuel b{ses ol menl.l imu8cry', in M.S. C{zznigd (Ed ), 7}? (ntnrn?N?!&ei?rc?J(CanrhridSc, MAr MII I'rc$).

Fi.rc, S. M. & Schooler. J W. (2002), II.wd you sel hero Irom rhcrc I vcfirl oveBhadowjn8 or sprr ir lmcntal fdch , A/t,lif./ (:atr iti r! ?D.halosr,16,9p.E91-910,

FEdrickson, B.L & Kahncmrn. r) ( I993), 'Durulion ncSlcot in rcmsnecrivc cldudtidrs of al l.c rivecpi\olcr' . ,humul of Paryonality aad Social hy.habtr,65. pO. 45-55,

qiubra, L.M. (1995). A hbo(nny mcthod lbr invcstiS{ting jnflucnces on swirohinS.rc, iin, r)tlsk-untl0led im{gc!, {nil $otqht,tunk i tv( \! thtl Catnlion.4, 00. l -21.

Ciambru, l . .M , Roscnbcry, E. l l . , K{sper. S . Ycc, w. & snck, l ) .A. (1988 89). A c i radnrn lhy(hn inthc licqucncy ofsnonldncoDs rNk{lnrcluEd imaScN lid rlrortglt\s . tnaNinatioft, Cqtnition, untt lt.renutrr, a, pp. 307-12

Crcdsky, A. & Cidhbft, L M. (lr1)(!)l). 'Thc orsiitorcy rcros visilMoc lnd ,cading l$ts 01 iftii!idualdiffcrcnccs ii thco.curen.c oftdsk{rnrclarcd andt.sk{cldled im{gcs unddrouShli, /,Lr!t&tbn, Cbkniti.n, an.lh,rioMlrv. 10, tp 19-52.

llilguld, E.R. (1969). lhii 11x prrlle li'r psy.hobgy an d phyrliali!! , A,ktiaa plrthrl.ri\t,21,DD l0l-13.

HilSxd, Elt. (19?ll), 'Colcn puiD in hynnolic xnnlgesir: Ils rculity !s tcsrcd by rhs red slmulurd<lcN)gn ,Journal ul Abnomul Piv./Dr,tl. ll7. pt.655-63

I ldNcr, 1 ., l.iorc, S.M &Schoolcf,J.W (2m3). 'Vc$al ovsslDdowin8 of mu \ic mcmory Whalhnppcn\ $ hcn ) .u dc,c lhe rh.r r ' rncr ' Unpdbl i \hrLl mrnu\Jaf l

Jx.k. A.l. (2004). Whar is nemiar.eness and why docs it mlrtcrl', Conllrcnc. prcscnr.tion, thcBrain ud Ih Scli Washinglon Unive6ily, Sr. Louis. MO.

lack, A.l. & R{rcpstodl A (2(,02a), lnfospeclion ud cognitivc brain m.pDing] Frum (tmutu{rcsponsc m script repDrl , n?nd! in Cognnire &i.ac.t,6,tt.333 L

Jack, A.l & Rocp(orfi A (2002b), The measuEnent prcblcm for cxnerience: DrmlgirA flaw .rinrriguing puzzlo? ,'/r.ndi ia C.tnitir( S<i?nus,6,pp 312-l

J.ck, A.l. & Shduice. T (2001 ). 'l nrhspccti!c physicdl ism as M approach nr dE rcicrce of conscbu\ne\f, a,snrn',, ?9, m. 161 96.

lames, w. (189o).Ili. Pfi,( i,/.r,/ Prr.n,l,rr (New Yo.t: Ildr).Kahneman. D.. Frc.l.ic kson. B. | ,., Schreiber, C.A. & Redelmeier. D.A { 1993), When dd.€ Dain is Drc

tetud ro less: Adding a beter enl , Plt.hnbln al:;(i.ncc,4, pp.4Ol-5.Krhncmai, D., Watker, P P. & Sarin. R. {1997). 'Back lo llcnlhrr I ExDloruLbns ofcxncriciced ulil

i r r ' , Quan ,b.hunaLtt t in , f ln l - l12, pt .375 1105.(c.gh. E. & Hedenl l ld l . M. (2002). Cender. cot ing and thc percepl ion olpr in. P! t .97.

Ko$lyn. S.|\,1. ( 1980). 1,r,!" d"l,'!ti,l (Crmbridge. MA: Ilrwd linivcsiry l)ress).Kosslyr. S M. ( l98l). Cft,rrr in nt( Min.l ! Mthinr (New \ork: W W Norron).

THE PARADOX OF INTROSPECTIUN

t8 J-W. SCHO{ 'LER & CA. SCHREIBER

Kr$l \n. \M .- . ( ( .8. t ' rnr 'D \ &\ ' i Cr( ' r f \ \ '4 lo6b' scur!r ' r ' l

, ,il i l ' i '; ',; l:i ;'; l,;;' j l :i '"1.1'i;".i:.' l; '., '^..,' 'c,,e..,.ro in,r'rc ,en(e ^ '|hcr'' ,.;: iiTl);;i, i;:il l:i";:ii ' i i l;.,' l:Ji'; '"1;.'::."{.,e rBD LJneil NxJc' 'r ""'.):tili; ;ii''j;,";l l;:":ll lil;l;iil i:'iirlll'i'. *,r",.*,. " "*'"e.,"- on,

,-#:[]il;lli:i:fiiii:!'i:li"d"u,u * t'o,ur. u-.,'n".a,,en,bn ,n'rrhesranrerencx Pr)'n'

"iai:.:"lii"il.:',4:#]i J'l,heqs, N.N. (re78)"tndividurrdirrcrcnces inrhe vcrir'|c'lriond sen

":i:;*ill'l;*:;:*;;l;r',9::{l;ltil;lul.tliiiSiif,'l'ilill;,i,1;Klli;l:l'",.,ii1l:111 i:J;"ij; l ' ' i l";i. ': ' l!.?il ':., ',,,,.,,,eir,,ucn',,,,e,r{\h.prnrc.c.in\e,bJlnveFl 'xd"{ .n! Mn u htdI r r r tu+ lu pp l - ' ' \6

"*tll;x**:;l;: ":x:11.ff i:lr:ff :lT:[i T,) ffi#;i;:';:'j".fi l:i,'J;i;::i"J,'iu3l!.Xi

*tr.t*,,.* (inrrcss), r'crceplurr lnd conccprtr3r cxperthc ne'riarc rh' lcrbrr ovcl

"illi'":llt-J';::ll"jl'lliiifil .iliJiil,'.flilill;ffi1,liliil'li'iil"'-,'. '* -,o,u'* o,*"-*'

"liill,:"iil'ijl1i;fil'{r",i"ifii1'"i'i1i;?.';,ii'l'i,l;i"1,,,. ,h0 s,,n or ,pue,rnsnli..4,r"i.d,

'i:l:";illll;',i,'ilii'ti,i,i:;,T;i,1'n];*,,, ,,,,^*, ,,*,3., n, ,, ̂ r. 4 /J/,{r} (Ncw rhvcn, Nri

,,",t;li !]ili:iillli.'i.cil! ".rir.

(:.M (2002). .vcrhdr ovc,rhudowiis in voice rccostri(irr', A lnrl

,,,j"i-i^"Jl"l,lii;llillriill:ill;"", ."...,.,".,,,n,{,\r,,{1,,,,c.r,nJ{*,d nc,Jrnf,,,i;::;irli',(,","li'iiiii;lll:1i'J;i'lii;lf;flil:-li(,..k., \1"q,.,,* /\./"/.,,*.|:i,ilJl:l'i,il'!",:;",'.'nun, ,) .,),6). ,,:rl,ll;.,,i:l:lli:l;:;:1,.'.illlJl X,r,i{.lli*:l'Rcdl litrrc dnd ro(ruslcoLivcevdLurlioii ol lwo min

-"ll,tiiljli:lir;i*mrilslltijtTlilt;i';lr,iffi,,H)l'jl:.il,,ll;i;;;,Y;1,'#'i'll;liij;l,1atiu h Ltcntut thdkctv tN.v Yddr oxlord univqllv(Pr.),iLr tn r,"q,""rt.,, .t-u L tu,ug*y,.r.h"*"i:iJli ,l;ll''Jilil:il"f;.1,,*,,iiilli.';Y"Ui-'.l;: ';;;,,r.ii.ri'ji"*, r,r,,rii.,r,.i,q,i,i*ir.r,r,,,i

/r j . 17, pP. l1t ( ,0.s. r, . . . , . j . i r ' ' ' " ,u .

AJJr. ! ! . ' \ ' i l . '

\ . : .nr . . r t '1\L1Fn' \ ' lu -{ ' \7i r i ' \ '1. . lW r , )ur '1 . Dr*ntcrnC rrc"urk^ 'Jbhc r ' l t l r r f r - l

..fji,'.'ll'i? [6i::lTii;.111,i,.J:ti:i"i:'; fi.:'l," ' vi,h in,rcsp.c,ion: ,.es*nse,o rrcr xnd

,"Ll'''".'l1'l; i'"d;'i;i l,ii1{i.""iiiL.."'i"1;.i';,i..'".l, 'is$ci,, .ns be,{.cn coiscbusncss nnd

dNta consciotr*esi, 7)?,4 i, c,tnrt!.t tr.r,6, tir lil]^ri", i,,q,r..p,tut" p.,."*i"e ,r,;t..Schooler, l.W (101)2c). Verbal oveAhldowng p&)

"lfll;lffi'Jrij#3'i}l"Trfrit-i!i:ul;r*;ul;iii.*::r::rli:,;i,:r'.Yr};;."l\illliliii'"'i'J;"",..,rsee), r'hesvn,hiosisot\ubjerlivcdnd.xFontrturrpp'|orhcstoin'iu

,.ilili";.'1'lilt{"'i{i; 13;iliiii.il'{1'rLii,lT;l;. "lershldown,s o, ! su! medo cs: somc,: l l : i " ' i f" i , i , ' i . ' ; ,"" i ; ; l l l 'J i i ' i ,J; ' ' ' ; ' ' '1"1i" i""" \ ' ,b ' ,di, ' - . , .N'-:'; : . ; ;

' . ' r , " ; : ' , ' ; . no. i v.rrn,rr , . i r ' , t " t t ' . J ' I1r ' r t4 &/M'r ' 'J ' rs"Drce'

CAr Acad.nri. I'rcs)S"t,,.l.r. r w., or'l*.". s n llrook!, K (1991), ThouShts b'v'nd wordsr wlcD lanEurgc oveNh&-,,*si".tigr,r',.r,.,-i,; r,2ttinkntdl P\rr holarv: t;oetul 122 ll 166 8:l

THE PARADOX OF INIROSPECTION 39

Schoolcr. J.w.. Rci.hlc. E.D. & tlalpem D ! (in pre$), Zon ing oul durnrg 'cadins: Ev nbncc lor dr

sociations bctscen e$cricnce and ncla conscroustrcss . in D.T l-cvin (Ed), ftintl'! d"d S"trsVisuul M&lcornition inAdults dn./ alild., (Camb.id.!e MA: MIT Fress)'

Schools. J.W. & lchrciber, C. A ( in pre$), loknowornottoknow:Consciousness dela 'onscbusnc-, ino nrotivtirn , iI t'org,". t.r., wittlu'"., K R. and von Hifl)cl, w. (EIls) ('ns'io$ andN.i Cxns(bus Matiwtbn (New vork: Canbrnlgc UniveAilv l'rcss)

s.nrrroer ( \ (2uul Pre, ihr . r 'er . r rn 'p. . r r 'cPlnh" lheuUnh lud menr uihir ' f rnpleJ'd ' rcf 'n ' lc 'Arr .mevor l f r 'dr l ler(nr i r :nErheel l i , r . iJrr ' r ion cmpUrl l l+srron Jn (onrrpurrrrerrure!I Jnpublished dissenrlion.

sdrclb{. C.A. & Kahncman, D. (2000). DerenniDanrs ol rhe rcmcmb€red urililv of aversivc soundsJormat of E p?rinental Prr. hobsr: ( hh. ral, 129, p| 2142.

Sclla6. W. ir963)..t i.,.., P! t..Ption atul R.alb ll ontlon: Roudcdc & Kesin l'aul)sheprrd, R.fl. & coop... LA. \t\82). M.nkLl thnsc! atuL 7,?ir Tidnrli'rtzrDtrr (cdrbridSc, MA

shepdl- R.N. & Mevlcr, J (t9?1). Menlrl rolalbn oi thrcc_dimcnsional ohjccls-, Jin'tu!. l7l'

siJilcr.i.s. & srcrn, E. (1998), conscious 3nd uncois.ious srfulesv dhco!cries: A mi'roscn€ric dnalisis , Joulnal of Everih.nlal

^!(holoB!: A41. nr| 127 pl 177-91

sl,iLer. J.1.. (198i). isamplins onloins aonsciounrcs dtrl cnNrbnrl expcficnce: lntrlislnns lorli alih . in M.r. Homwirz (E,l .), tirrtAbno"i.j aML (i{irn r (chi.r8o. l L: Univcnirv of chrcrso

sn[llw;od, J.M.. Bdfuclia, s.F., L.!wc. M & obrrsrwin. M (200:]). Trsl utrrclatcd lhought whil(cncodin! inlbrnrlttun . (iJnr.k,6n(i anrl Cot<ri|idt.12'Fp 452-81

Smal lwooi . l M.. Obonsawir , M.C &l lc i f t ,St) . (2003). 1rsk unrclnled l lnughl : The roleoidi(nburclorn.c".r ! , t1n\1 ' , ' , . \a"J( fa," ' 12 f r l ' ( ) l { ' )

sr . . ' l l r ; , ,1.J.M.r)c.nrJr,RC..5r h. j M\ ^ ()1. ' r ' ' ' r i t V :nn4l l '+ un'cLr le r f 'Lf l ' l

;nd dvsDhorir: BchdviflnrL!nd physid;8ic,Ll .onrcqoeno.s durins enoo(litr8 t'npublislrcd nldn!*IDl : ! , I r '$ t " . (

l , r r l t r r ' \cr .JJ . hr ' l - r I r , rc^ i ) fSuJr| ' | lJJJ\ ! r r l l ; . ! , r i . i l !nt , . l l 'c In l ,nnir , iJrr ' l 'h lc I ' bn' i ! r 'u.J rr . ' r r ' i icn ' - l t -hr t t " r tMrr"

xr u,r \ ,74, cnt i re i$tr . .sr ' ,n ln r t . . Socn.rr . s.r & Z ' r t r ' . M.f 2 ' { ' : ' srblr ' ru l f r 'mnr! i ld Ic.u.^ n \ r r rhn!

"h ' 'lhc non i\ hoi , ./,r, i?/ ?y' tr rltihvnd soci?l t'\t loloNt 38. m 556-68'

sr r .AA.,&odcr ict ,J. l i . ,K! ; l l .Al .Dclc: l )d.PA.E0&Porrcr, l - .1-"(2lX)0),Docsrhcl lcrk- ' r lrrhonrnncnor obsowslin labtnrlrtry lrh sludics rpDlv rr) Icrl world frii ir tumrkrd r hri(icsl't tutrr l , ' l l \ r t , L pP l l l l1

l . . r , : . 1 i , l t r . , r i : l l , l l .1D'" MJ 11 "n{ l l ' ) .1 (a \ rnr ' 's f l r \ l - ' r ! / ' l { r5 '' \ r i1r . ln. 1J. | . t r , l rn ' ln rrJr . lcNl l ' . ,n (Lnrr ' i r \ (ur i \

v" , i t , Lr , t "n,r , r"u, ' r ' t ' .n r . f f ' l !J / x, ' " / " { r ' rnwr\ \ ' ( r ' l r ! \ 'J '4r 'nr_1ln-1r)$.1. , r ' , l . la t ' , i .P, \ , / . , , / , t t t t , r , t \ ' : t , r , , l . '4t , - , t l r , r ' , 4 r 'Ph . f t l l f i - l l \ ' I pr ' rJfrWelrcr . D.M. { l !9 i l ) . l ro i icpf t ) .esecsolnrcr l i lcdn(nn,/ '$d! ' ] / ' * iz l lRd!t , l t ' l , l l1452wir l . " . rD..r i . ' ,bs. . l ryb;c,J.A. l l lmr. l ) l l &l lnondo, lA (1e84) El lcer\ofunr lvzinsrc!

*ms on xll itude'bcl[vitr ;drri(cicl l o ill .l hr soialirv rnd S'clal P\vchoiogy -:l? 5' 1 6wi l$n, l . l ) . L indsc! , S. &Schoolcf . i .Y ( :000), A d!rd0l o ldur ldui tudcs , / ' \ r r / r ' l , r i ' / / / i ' rn ' ' '

107. Dp. l0 l 2 l 'W'r"rnl r .O.. t - i .1. , t ) .1. , S.hoolcr . JW,l l .dgcs S l ) . Kl re i K.J & Ltr i lcur. -SJ' (19' ' l ) '

'lntfospeciinA.bour rcannrs cxi rcducepon choice silisla!ri1tn .l'auhaln! anlSo(nn PsrthobsrBu|1l in, l t t , l l l 9.

wi i . . i i r ' .b as"nn.t" , , . t .w.(1991), ' lh i t r [ insl [ ]much:Inrrosp.ctn)nc"Nducerhequnl i tvofprel.rcnces r id deci\ioi s l . ,,, nol 4 fe&,vlrr .nl ttu ial Pr)

'/r,/.']s1. 6{', PP lSl 92