Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.psy.msu.ru/science/public/krinchik/krinchik1974.pdf Дата изменения: Mon Oct 22 00:20:44 2007 Дата индексирования: Tue Oct 2 05:31:40 2012 Кодировка:

f

Perc ption & P shoph sis 1974,VoL. I5,No. 1, 131-144

Probabili effets in hoie reation time tsks
LNA KRINIiIK*
Mosol State Univrsit, Lenin Hills, Morcow V.234' U.S.S.R. This paper ontains a short-reew of the main rsults tht wre obtaind by th autlror in a seris of experimnts that onstituted study of th ffts of sl pobability on hoi reatin tim. Th effts of stimulus pobaility ar slrown to b influend by the follong variables: (l) diffeensin the method of varying stimulus pobability, (2) differensi task ompleity' (3) differenes S.R ode, and (4) differenes Ss'motivation.h dt that aI in in onsideed hr re the overall mn RT for partiulr signals and th me RT for squntil rpetitions. Two ..nature'' of qustions, reltd t the psyholoal the pobability ffts in hoi RT a disussd:(l) Th qustion of the relationship btwen th relativ frquny and th numbr of ltrntivesas two diffrnt ways of determining th plobability effet i hoi RT; and (2) the question of identifying the main determinnts of the tril-to-tril vaibility of RT in suh exprinnts.

Th first efforts to pply probbilisti apprah to the analysis of animal nd humn bhavior we mad sevral deads ago (Brunsk, 1928, |939, |94; Romanov' l935; Skinnr, 1938). So muh dta and so many modls using this approah hv aumrrlatd sin thn that branh of psyhlogyhs now mrgd tht speializs in th study of stohasti aspets of bhavio.Within this ara, a speialplae is oupied by th invstigation of man's behavior in probabilisti situations' suh as the analysis of probbility larning, the ffets of stimulus Probability in rtion tim (RT), t. This pape attmpts to prsnt systmatianalysis of som aspts of th probabilit fft in hi RT priments and to fomlate hypothsis onrning its main dtrminants.

oD
Since the results of a lage sries of eperiments will be considred in this papr, th pimentl method will b desribd i a gnera.lway at fist. The mthodologicl details petaining t a pticular expriment will then be ven at the time the results for tht xpeimentar repoted. Th prinipal rsults we obtaind by using the typs of expimnts:(1) xperimentswitlt two.oie tsks,whr the indeendnt rible was the relativ frqueny of th stimrrli; (2) epiments with multiple.hoi tasks, whr th indendnt vaiablewas the umber of equiprobble altelntive signals;and (3) eperimnts with multiple-hoi tsks whe the two idependent vaibls wel the numbr of ltnativsand thir rlatirrcfequenis. ah piment begn with a prtie sssion,in whih th simple RT and the hoi RT wele measutedwith equiprobable sals. Sals were prsnted ndomly aording to a speil program' th the intertlial intvl (ITI) set betwen 5 and l0 s' depndig on tie epimet. In those exprimnts wh the ITI edd 5 s, a wring signl ws prsented l.5-2.0 s bfor the test slcnal. Varius kinds of stimulus-spons ods wr used: lights to kys @igs. 2 and 2F)' lights to vrbl symbols (Fics. lA nd 1B' and Figs' 2G to *I would like to xprss my dep gatitude to Profssor A. N. ws startrd Iontjv, on whose inititive this srisof rserhs svrl years ago, nd under whose dit guidn th st of th studis dsribd hr was prformd. I am lso spei2|ly indebted to Sylvan Konlum fol his dtiled ommnts on preus drafts of this paper. sver!

2I), geomtri figurs to keys (Fig. 1C), sounds to kys (Figs. 2A nd 2B)' two{igit numbes to thir verb] names (F!c. 2)' twdigit numbers to kys (ables 3 ad 4). When lts were used as stimuli' tlty wer ommuttor's lmps (24 v, 0.105 A, 10 mm in diam). ln th two.oie eperiments, they we plaed hoizontally on a vertil pael, w.ith a distane betwen thei enters of 75 mm (Leontjev & rinhik' l962; Krinhik, dnikrov,& Borisv,l970). In th multipl-hie experiments, these lamps wee plad on a vrtial panl i a 3 3 matrix, with a distan btwn entrrs of tlre two net lmps of 25 mm (Leontjev & Iftinhik' 1964). Geometril figures we dawn in bik on white ds.Their siz ws 20 2o mm. They had contousof a ure, a irl,a triangl, a staf, ross, th s of infiity, and spirl. Thy wr presented in th window (95 50 mm) of a thistsope (Krihil & Alexandrova' 1966). When numbers w used as stimrrli, slids with blk numb insibed in whit uIes were psntedon th sIen with an pidiasop (Krinhik & Akimova, 1970; I0inhik & leandova, 1970). Th pojeted size of h numbr rvas equal to 30 x 15 mm, and of th whit squae' l00 x 100 mm. Dt fo the sounds usd as stimuli Ie prsentedin Setio B. Whn kypss (or buttonpress) rtions we used, th eativ kys (or button) were pld either on the horizontl panel rrangdon th table bfor S ol on th rmhai. Verbal leations wer registeedthrough the mirophone or lrigophone and with spil uipmt whih stoPPrd ti tim t th stat of th vrbl eation. In most epei.rnents,RT ws msud with n eletroni tim'SK-l' in 0.001 mse. Depnding on th peiment, th length of a stimrrlus sequnrraidbetwen 80 and 240 tils. ln th s of shot squens' Ss mit be rrn up to two or tlrree sequenespr dy. In the main sssios, th squs hving th sam stimrrlus pobability distributions were replited from two to eight tims, dpending on th eperiment. ln all th expeiments, th S ws rquird to tet s fast as possible thout making eos.Data of ol te, as wll s sttistis of trils per dta point, taining pr S' and othr details of th erimntal methods are prsented in Tabls l nd 2. rronous Rs were not inludd in th ta. During the eprimnt, the S ws not info'rmd bout th probabilisti stlutur of the sequn' nor was h givn fdbak on his speed or aay' ept i those eperimnts where differential reinfomnt ws the viable (s Fi' 2F.2l). In e lttr as,qpel mthods of rifomt wer usd when Ss xeeded the predetmined tim dedline ding the urent urse of th sssion: .quipmnt brkdown'' reinforment in the two-oi tsk [se xplanations in ) blow] , ad the diffrnt types of riformnt (rd lit' shok' nd ..equipmntakdown) in

131


|2

KRINHIK
Trils Pr Dt oit and Tbl l Rt for eh primt. or

rimnt Fig. 1A ontjev & Kihik' 1964

Conditions

Stimulus Probability

Trils Pel Data Point

rror Rate Per Dat Point (in Pecent)
Totl Per S

Tw.hoi tsk Lits to Verbal Symbols IntrtrilInterval(ITI)- l 0 Se Forpriod(FP)_2 S Warnig Sit (wS)-Bll Tlials Pr Days Session N=240 Ss=6 ultipleChoi Task With quiprobabl Sials Lits to Verbal Symbols ITI-l0 S F-2 S WS-Bll N=240 Ss=6 ultipl*Choi Task With Nonquiprobbl Signls n = 2 Gometial Figus to Keys n=4 lTI_7 Se FP_2 Se WS-Bll N=120;n=8 Ss=6 Dondrs'sA' C' and B Rtion ask (Without Waming Signls) Soundsto Keys ITI_5 S N = 120 Ss=5

0.07 0.s0 0.93

'120 5760 r0800

3.s 6 0.39 0.1 3

0.59 0.07 0.02

Fig. lB (UpPr Curv) Leotjv& Kinhik' l964 (Dat for All Alternativ Sienls)

25 0.1 0.2s 0.5

1440 1440 1440

0.62 0.9'7 0.62

0.10 0.l6 0.10

Fig. lz Krinhik & Alendov,l966 (Data for the ritial Signl)

0.07 0.50 0.93 0.0? 0.50 3 0.9 0.07 0.50 0.93 A Rtions 0.07 0.50 0.93 C Reations 0.07 0.50 0.9 B Rtions 0.07 0.5 0 0.9 A Reations 0.0? 0.50 0.93 C Ratins 0.07 0.s0 0.93 B Rations 0.07 0.50 0.93 0.07 0.50 0.93 0.0? 0.50 0.9 0.07 0.50 0.93

1440 2688 2688 192 1440 2688

r9z

r92 t440

9.90 1.59 0.1 I 2.00 0.5 0 0.15 2.00 0.90 0.07

1.65 0.2'l 0.02 0.33 0.08 0.03 0.33 0.l5 0.01

Fig.2 Kinhik & Rajvskij'l970

128 1800 1080 136 2280 1904 l0 2400 2240 64 960 896 4 960 1008 96 960 lt20 96 720 1340 96

@at for th Critial Sil)

t.4'l 0.61 0.42 7.5 0 0.5 8 0.22

0.29 0.12 0.08 1.87 0.l5 0.06

Fig. 2B ihik & Rjevskij' l970 (Dat for the CitiI Signal)

Dodrs'sA' C' and B Retion Tsks (With Wamig Signl) Soundsto KeYs ITI_5 S P_l.5 S WS-on of l000 tlz N = 120 Ss=4

0 1.04 0.88 0.94 0.10 0 3 8.3 0.69 0 .|2 0.9 0 1.04 0.21 0

0 0.26 0.22 0.24 0.03 0 |,9 0.12 0 0.52 0 0.26 0.05 0

Ft.23 Krinhft & Akimova' l9?0 (Da for thitil si8n.l)

Multipl{hoie Task \Uith Nonquiprobabl Sign n = 2 Numbrsto Kys lTI_10 Se FP_2Se n=4 WS-Bell N=120;n=8 Ss=6

'r20 1340
96 't20 I 340

o.tz


PROBABILITYF.FTSIN HoI RATloN TI TAsKs
Tal l Continued Stimulus Probability Tials Per Dat Point

l3

rror Rat Dat Point (in Prent) Totl Prr S

pIiment abl 3 Krinhik & Akimova. l970 o)at fo th Critial signl)

onditions

Multiple{hoie asks With NonquiProbabl Signals n = 2 Numbersto eys ITI_10 S n=4 FP*2 S WS-Bell N=120;n=8;Ss=6 Solving DivisionProblen-r ITI_10 S FP -2 S WS -Bll N= 120 Ss=9

0.07 0.50 0.93 0.07 0.50 3 0.9 0.07 0 0.5 0.93 0.'7 0.50 0.9 0.07 0 0.5 0.93 0.0? 0.s0 0.93 0.o't 0 0.5 0.93 0.07 0.50 0.93 0.07 0.50 0.93 0.07 0 0.5 0.93 0.07 0.50 0.9 25 0.1 0.25 0.5 0 25 0.1 0.25 0.50 0.125 0.25 0.50 0.125 0.2s 0.50 25 0.1 0.25 0.s0 0.125 0.25 0.50

Fig. 2D rinhjk & Alexdrova,i970

Positiv Responses 5.14 408 0.64 240 0.5 9 5't12

10.41 96 2.22 720 0'29 l40 96 r 0.41 0.83 120 0 1340 9..| 96 0.83 720 0 |40 Responses Ngativ 5.88 408 1,02 240 0.99 5112

|.1 0..7 0.0s |..7 0.14 0 1.56 0.14 0 0.5 I 0.1 I 0.1 7 0.5 0.t)'l 0.01 0'29 0.03 0.9 0.06 0.02 0.0? 0.04 0.01
0.65

Fi,' 2a Kinhik' ednikarov, & Borisova. 1970

Two{hoi sk With Intllavd DistatorTask Lits t Kys II-10 S ol a threperitnts N= 180 Ss= 13

156

xpimentl

,84

156 5l0

2184 0,36 Il pimnt

0.2'1 2t84 peimnt III 0.92 1s6 0.55 2340 0.18 2184
Bakground periment 1.96 255 2400 0.7 0.02 75 Reinford Sal 45 2.22 360 0 0 675 Nonrinford Signal 45 2.22 l.6 360 675 0.59

5.12 0.1't

Fig. 2F Leontjev& Krinhik' 1962 (Data fo th Iitial signl)

whoie Task peIimntsWithout Ri. foremntand With qui ment "Breakdown" Riforemnt Lights to Keys ITI-l0 S FP-2 S WS-Bell N=240 Ss=3 Multiplehoi Tsk.With quiprobablSignalsand Diffent Typs of Rinfment Lits to Vrbal Symbols ITI-10 Se F_2 S WS-Bell N=240 Ss=6

0.r2

0.006 0.'14 0 0 0.'14 0.5s 0.19 0.09 0.2 0.t2 0.0? 0.46 0.18 0.0s 0.18

Fig.2GIs Lontjev& rihft' 1964 (Dta for th Citit Signal)

Fig. 2G: BakgroudSessions 180 O.ss l.94 360 '120 0.69 ReiforedSignal 18000 240 0.41 24000 Fig. 2H: BakgroundSssions 2.',1'l r80 l.ll 360 0.2'I 720 RinfodSignal 1.1 I 180 24000 24000 Fig. 2l: Bkground Sessions 18000 24000 0.14 720 RinfoedSil r.6s r80 24000 24000

0.02 0.28


|4

KRINHI
Tal l Cotinud Stimulus Probability ials Per Dat Point
Noneinfod

rror Rat Pr Data Point (in Pcnt)
Total Signl Pr S

priment Leontjev & Krinhik' l964 (Dat for Nonreinfored Sils)

onditions

Se Preedig Setion

25 0.1 0.25 0.5 0

8s0 680 80

2.82 |.r'l 0.44

0.47 0.20 0.07

'Trials pr dt point nd rror rt do not inlud the dt of th trinin sessioris.rror rat dat r givn in t\)o vaints: s a totl nor rt pr dto point nd s n vrg rror rot pr S. 2Dt for th citil signl ar th vrg dt for two signals thot wr presentd smmetrioll in ll th thre st siz conditiotls (N = 2' 4,8). Two diffrnt figurs wr usd s citicl signls for diffrnt Ss. Hr onl on itil sigal wos usd nd it ws diffret for ah of th thr st siz onditios. But ths thre diffrnt citical sigls remind th sam for ll Ss i ths prits. aah probabilit condition hs 90 trials. Two stth conditios wer usd i 1 da's sesion. 5ll three st siz conditions wer used in ech d's sssion in th bckgtound sssiltts wll s in rinfrcd sesstons. os Total trials pr d's sssion wr qul to 240 (80 triIs pr h st siz condition). the multil-hoi task Iseeexplntions (D) blow]. in At the stIt of the sessionswith reinforemnt,th Ss wre infomdof th time dadljn..It was different for eh S in th two.hoitask and was ual to th fastestman RT strownby this S in th bakground sssions(i.., th sssionswithout einforement) with equipobabl signls.In th mrrltipl-hoi t.sk, time dadlinewas equal to 400 msecfo ll Ss. th wo sssions with reiormntwr ondutd i th two-hoie periment: on with the reinfoed rre signI,d th otie rvith the inforedfquent signl..Inboth ases, the same signl was riford. The sssions th reinfomnt wr ondutdin th multiple-hoi epeiment.A-lltheest size conditions (N = 2' 4, 8) w prsentedin 1 day's sessio, and the of eight signls.wee reifoled with tfuee diffrent typesof reinfoement. Training reinformnt sssions peded ti main sessions with einfomet both periments'In tie two-hoitsk, in the wre two suh sssionsrvith equiprobabl signalsin whih the red light flashdwhen Ss exededth time ddlie.In the multiple-hoi task, the was o training sssion i whilr thre of eightsignals(differentfom the rinfordsils in the main sessions) were einforedwith th tone when Ss eedd the tim deadlin. ai dt fom the reinformntondition wre omprd in all ses with wellaatied dta from the bkground ondition, whih Jwayspdedthe reinformntondition d whih onsisted of my sessions(seeTables I and 2). e Ss wer studnts from diffent depaltments at osow state Univrrsity. Thy wee niv about the purposes of the experiments,but some of them took palt i sevealof th exprimnts. Additionl hngesin primntlonditions onenedthe method in whih stimulus proability was vrid, th hrate of th exrimental task, and the type of inforcement that was used.
h Averag able 2 Amount of Training Per Subjt Average Stimulus Amount of Poba. Taining bility Tials Pe S

Figures Fig. 1A Fig. I B (Upp urv) Fig. l F-ig.2A F'ig.28 Fig. 2C

i 0'50 1 0.1l-s 0.ls 0.5 0 l 25 0.1 1' 0,50 I 0 0,5 0,125 t able 3: Numbers to Ks N=2 N=4 N=8 0,125 n )5 0,50

1560 260 1360 786 786 786 960 1560 42 1032 360 390 520 48 90 160 6s 0 <{) 520

Fig.2D Fig. 2 Solving a Division Poblem rtg. tr Bakgound Sssions Reinford Sssion

r 0,50
0,5 0
l 0,50 0.50 I

333 600
240 26 180 960 I 840 480

RsULTS (A) The ffts f Diffrent Mthods of Yarying Stimulus Probabilit Figur\A' prsnts th su\ts ot th tw-ho! xpimnt in whih stimulus pobabi\\ WaS vand b (Iontjev hanging the relative frqueny of the sign-ls & Krinhik, 1964). stimulus probbilities rngd frm 0.07 to 0.9 and wr ountrblaned both sials. for Undr these onditions' and with th anse of sisnal
Fig. 2.l

BakgloundSssions

RrtordSssions

I 15600 0,12s 680 0,25 680 0,50 680 0,\2s 80 0').s B0 0.50 80

probbilities that was used, th fft of probability on overall RT apparsto b satisfatoilypprimatdb strait lin' Th similitybetwnthsersultsnd


PROBAILITY FFTS IN Gol RAION TI AsKs

l5

thos obtained in othr studis is stiking (Iontjv & Krinhik, 1962; krte|son & Barzl, 1965; lming, l969; Rmington' l969). Differensin th interpt ar appIntly due to differens in some partiular onditionsof the xperimnt. t I In Fig. lB, the ffet of stimulus probabilit i a I two.hoi task is ompard w.ith th ft f stinulus probabiiity whn it is varid by hangingth umbr of quiprobble sials tim two, to four, to it (I.ontjv& rinhik, l964). Both prrmnts wr ondutedwith th smss and th smequipment. he dt suggst tht thesetwo dirnt mthdsof rir) varying stiruls pobbilit.ompris two dirr;t f oionnts in th robaiilit;'stlutur ot. stimuIus squns. Th stdealswith tir rltiv fquny f r*'f fi-xed numbr o sigrraisnd will b dnoted s th ..rqtrnr.'' omponnt; th sond ,lis with difer;nesin th nunrber o ltrntivs and will he ,.', ..altrnativ'' drn'ied^s th mPonel]t. ln Fig. i r plottd th dta of n eeiJnnt in whih bth th nrrmbro .!irntivs th rjt;v nd rqunyo th signlswer hngdsimiri.lanusly ! ': ! i{i66). lrr ah f a iw., 1rinirik & A]xndr'v, lt.i. |our.' d .it-hi.-' lI tsk. ih "rbbi.litl, lln f. ,;hangd thc sials, th itial signi,ws trr .G,/'t .5' to .9.' th remining signals wie prsntd {r' lequiprobably.Th data in ig l r r th ritial sil only nd inditr th s.n f lr '''ii intertio btwen roblbility nd th numb of , ,'j,, altrnt,ivs' Sirilar sults lvr lbtind by i'r}rik d kimor'(l970) (s:rhls iitd.4. the ntrrlrbrs 'l t ks task). Kornblum {|96.7 l969,|;isi.lan.iyzd relatinsirip ihc ' Fig i. (') wllt;ie d; th nuiprobl signis.Th th numbr f hsissrrq{nls stimuls pbilit' th olinte rrFsnts btwttstimu]us rtlb.rh;!it.anc ]trntivesin situatln ill whih h vid t}r ovrall m RT. h sttttli usd wr llpq d tlre sqg-nss vrbI..-.. lntjv & Ikinhitq l}fl| (l{f = probabilitis f rptitr'ns and nntitions t}r r-83p); t. +. -. Ben, J965 (R = 44.8-55p' F-P = 3an sil in two., fur.. rd it.hoi quiprobabl 5 s);e. -. r Brtelson' 1965 R: " 428 '|22p' FP = li2 s); tsk. His tesultswr quite similit ours, both f the ')*-o = 466 -7'lp); .--. Lernig. l96i ET rcPated nd for the nonpetedsials. Howvr, h Remiton, 1969 (RT = 32j -7). (B) ovrail mn RT for r found that the repetitions and nonpetitions wer two+hoie tsk with nonquipbh|e sigh --._- (Rf = _sopt d fo twr>, fout-, nd it.oi tek i whidr diffrntially ffetedby th number of altmtivsand, 6 ^ (R = E97 _lBEp) th the sigrls were quipobeble A in ompisn to our dat, were ls ftdby th stimuli onsistd of lnps nd e rco!Es wr vql signJpobability. (.eontjev & dnh 196.4). he dt er for on ptiul Sternbrg (l969), who also studid an intration signd tht tvs usd i oth primnts () ovrll me RT { = btwen stimurl probability and th numbe of for tlre ritil Fds i a twooig 3-1 3 -8), 194igg-r-^ fi = ,o -l92p), ad lterntivesin a biry lassifitintask, finds that th (R = 5l2 -2P\ t; th stimuli wr eighth l-. slop is affeted by the pobility. whras th gomtri urs nd the rsposes wl kyprsses (rinhik & intrptis ffetdby th numbr of a]trnativs. Alndrov. 196).
!

"'f
i

I

I

n RTs to the ritirt Stirnufus for Diffnt N.ff:::

3 of Altrtives

rd for Dilfnt

Stimuts Proilitis

Stimlus Pobabilitis Two Alterntives Sfud Krinhft & Aleandrova(l96) Kinhfl< Akimova (l970) & ask GomtiFiguresto Keys Numbsto ys Numbrsto Nms 0.07 31 5l9 540 0.5 25 45 485 Fu Altemtives 0.5 t1 481 48 438 602 530 2.I8 4 384 -g Al!"fi:s 0.5 0.93 49 667 622 96 285 525 66 4,I,| 389

0.93 0.07 268 34l 99

9.9 0.0?


16

KRINIK
Tl 4

Pobility ffet (Slope and Interpt) for perints With Diffrent Deee of Choie (Dt fo Critil Sial)
Number of lter. nativs ask Pramete

Krinhik & Alaadrova1966 Gomti Figures to Keys 4 460 512 -80 ,192 *240

Krinhik & Akimova' 10 Numbers to Keys 535
o.! I

ihik & Akimova' l970 Numbers to Nmes 552 550 540 -163 - 170 - 155

Kornblum.1969
Rpeated StCnal

Nonrpatd Sial 22.5 73. 450.5 -65.2 -19. - 103.7

Intrp1

2 2

Slope

694 -207 -300 _49

20.6 328.3 68.2 - 105.2 -84.6 - 110.5

(B) Probabilit ffets at Diffent Stagesof Informtion roessing Krinhik and Rajevskij (1970) net studiedtlr ffets of stirtilus probabilit in two perimntsin wlrih Dondrs'A. B, and tsks wr usd in odr to olnre th probebility ffet in a task rprsnting difrnt stags of inforrnatiotr Prossing. alr xperimentusd a difrent gloup of Ss, xpt for th B and tks perimntI, no warning signal was usd in this study. Th stimuli in primntI wr tons of 400 and 800 Hz. h stirnliin xprimentII wre tons o 250 nd 600 z' he stirnuli wr .f presented with pobbilitis .07' .5. nd .93 in il ondition O ah epeimn. W}rrtli task ws silpl R tsk' one o th ttnswas simly Omittd from tlre randomsequenes wirihhd prpard or one ol the oth.r tsks: in this wy. th tmporl distribution the solitrysignJiws omprabi ils f to distibutionwith on mor signal he sults. . wlriil are prsntd F.igs. and ]B, indiattht th effet rn ] probabilityWaSquil simi-lar th A nd B tesks.This in finding suggsts tlrat tl-re ftto robbrlityis losely riatd to tmpoalstruturs(..tirnemophology'')of th stimulus sequenes(Kinhik. l969: Krinhik & ednikrov. l970). Send' th warningsignaidid not hav an fft on the slope of th urv in th B task. This is in ontrast to Bertlsn and Brzl(1965)' whos findings imply tht' with the warning signlating as a u' varying th lngth of th forpriodwould fftthe slop (Fig. lA) of th line. Fin.lly' probability appers to have had a pulia fton th C ration.In both expimnts, data the for th C task dpart from linarity by wht apprsto b th prolongedRT in the equiprobbl situation'This indiates th importan of signal disrimination proesss th .ration task, sin ths prosss in mr't b mor ativ in the quiprobablsitution bus o the frequent lterntion of tlre signal.r orov. the urv for th C task rveals a lar tndenyto hangeits positionin th ordinateaording t th absnoI prsnof th warning signai. ln the formr ase,tlr RT in gneralis lser to th B.reation dat; in th latters.it is losrto th A.rtiondat.

(C) Probailitffts with Differnesin ask ompIity FigurJ illustrtes data tr tlre ritialsigni th of a study in whih oth tlr numbr o elterntivs nd stilnulus probbiiity wer vried tn tlutnbr-namig task (Krinhik & Akimov. l970). ln tiris study' vo-digit numbcrswee usd3s stjlnuli.witlr dierent number bing usd as the ritial stirnulusat al-t lvl '.]7'. o tlur.hoie, o hoie (..23'' r tw.lri. ..]5'. and or it-lroitsks). stimuiusprobabilit ertsan et lll dgres t o hoi;this fft also appears be tnoronound to (Fig. lC and ble.1 Howvr. than in previous studies ). tlr numbr o altrnativs dos nt smttt huvehd tl ffet in the numbr.naming task.rr.hreas did haven it fft in th kyprssingtask (abl.'s.] .rlrd.1). Btli -periments ondu1d wer rr,'itil sltlllt'srLlus Ss. l]lt' ol and th samestimulirvr ust.d. is dit'fiult S\l It tO now whethe tl-re tlsneoi n alirnliir' et.t du t is S-R omptibilitytlrtorstlt. tltis urltbr.r.naming tsk (ivra. [9)60) to th tlll1nrilrirrit two-digit or tli. nurbrs whih ld lO tile e]\.tlrln R for two. nd oi our.hi tasks (Fitts & Srvitzr.l962). In nY ase. ths dta on1lrrni!,lowbrav's results(1960) and in a diffrnt situatin prove tht the numbr.naming ode dos not abolish th et tiqueny onrponntof o stiru]us squn. owbray. too. hs obsrvd Ieriy epressed et of a requeny omporrnt in an periment rvith a numbr.naming tsk using sltiv rationand a diferent numbrof equiprobabl stimuli: fr N = ]. 4. 8. it lrad an intrepto549 and a slopof
11

otnprison of th dat btaind in the number.nming task and in th numbekypressing task supports Strnbrg's finding (1969) that tlrr is difrntloalization of probability and set size ffets on inormationprssing stgs.orovr,this rsult appars to suggest tht the probability fft is onnetdnot only with rsponse bias prosss t1re and rsponseorganiztionstag,or the invstigtd tasks equirsvry different responsorganization prosss, wlrrasth pobability etws signifiantly grt in both tasks.


PROBABILITY FFTS IN ttol RATION TI TAsKs

|3.7

Fig; 2. (A) RT s a funtion of stimlus probbility. Th stimuli wre tons (400 nd 800 tlz) and th rsponses wr keysses. Th resrIts e shown for tfu 6flf = tsks: f}onds' p A 122L-79p)' Donds'y B e_. _. -. [R = 36l _ll)' nd Dondrs'TpC l_ - - -l. Th .'ta r for on rtiutr sigrl whih ws usrd i h of the thr tasks @pimerlt l' Kirhik & Rjvskij' 1970). (B) R as fution f stirnulus probbilit. stiluli wr tones (250 d 600 tlz) arrd th rsposeswre keypresses. h lerl r sltow for tfu tsks: = Dondrs' p A .___. G 2!9--7oP), Dodrs'Ty B r.. ' -. -.A (RT = 322 _ll5P)' nd f}ondls'pC I & f_ _ _ -r. The dt I fr on prtiulr signl whih ws used i eh of the thre tsks. (riment lI' irrchik & Rjvskij' 1970). () ovtl m RT f th critil stimuti in numr.mi tsks whr th numbr of ltrtives and stimulus proililv wele vri_ sirultnous!. (R = 552 _l63p)' whie ..--. Aftrr-hoie _ .- _^ (R = 550 _l70P)' r and eit.hoi l- . - . _- .. (R = 540 _t55p) (rihik & kfuov"l9?0). t\ @) Rsps !ten in soh.,ira dsion problem. s t(t for eplartbn. l.tny fr positiv set . _.._.-. (R = l.39 - l72p), lten for ngativ st A_ - - _^ (RT = l.7l -407p) {'inhik & r Alendrov' l970)' ! () e R dte for nnuibbl twhoi tsks i whih th stimli wr lamps d th rsposs wae kprsses; with distrtor task - - _ _r (RT = 4!5 _|2p' pri,mattl)' A- . - . - .A @T = 420 -63p' priment l),:d (RT = w.ithout e disttol .e(k .-. 349 -61p' primt lI). h R for th quiprabl dition yith th distretq r trsk iditd b r (rihik' Mdikrov, & Boov 1970). *qumnt (F) ovrll m RT as funtio of stimulus probbility i t]r twohoi onequilioabl sk with th (R R for reiford signl A--------------^ = 245 -{3p)' RT for th sig'd i" kground rpeiment rkdown'' reinformet' = ._-.--. 1 = 329 -llsp), R for the nonreiford sig'al in the epeiment with reinfmt o- _ - -o (R 320 _99p). Stimuli wr lits d lspnsrs wee kypresss(Lontjev & rinhilq 1962). (-I) ovll mn RT s fitntion of stimr,lus proili for two, four., ad eit.hoi equiprobble tsks with diffrentil rinforments on s{td signls for tespose times dig 40o ms. Th RT to rinford sigrls idirtd s -' ad to th sm signl in the bekground epiment, s o- _ _ _.. Th stimuli wre lemps, th resposeswrvf,bl symbols Gonligy & ihik, 1964). (G) Reinforemnt onsisted of rd lt going on for ssivl log RTs With rinforernrrt,frT = 585 -58p, thout reinforrnnt,R = 733 _7l6p. () Rinforemt osted of l slrok be en aftr essively lo RTs. With rorert, RT = 376 -24op, thot rinformen! R = 768 _s36p. (l) Rinfqmrt osistd of the equipmrt .rk down'' aftr ssivd long RTs. rr{ith rinforement' R = 352 -l88p' without rinforement' R = 699 -676p. Fire 2D shows th fft of stimr.rlusprobability on overall mean RT for the solution of a simple aithmeti problem (IGinhik & Alexndrova, 1970). h task onsistd of detemining whether a thr.digit numbI was or ws not divisibl by 3 @odrov, Gnkin, & Zarakovski, 1966). The numbers were presntd on a sren with an Pidiasop, and were tuned off by th .!es'' ..no'' rspons. The Probability of or S's vrbl

T a ts


l38

KRINCHI

onsisting th division problmdsribd of abov.Two division problemswr prsntd during the l0.s ITI of th major task (Krinlrik, dnirov, & Borisova, |9'70). Th study onsistd of th ollowing Sssions With th with th Sessins xprimnts:pratie sessions, o mesuremnt simple DistratoTask Bfor Distratorsk Ate nd lroi RT (two.hoie quiprobabl task); g B'.k9gj:1e, ! J"!9!'].s xperimentI, two.hoie RT with nonequiprobabl Peen1 eent Percnt Pnt stimuii nd distrto task; xprimntII, two-hoi rrors omissions ros onrissions RT with nonquiprobbl stimuli and without distrator 0.5 9.2 10.5 task; xprimntIII. tlr sm s xprimntI. In ll 0.07' 4.1 0.9 8.1 6.6 5.3 thsprimnts, samset o stimuiusprobabilities th was used.An quiprobabl onditinwith th distrator negativeand positive instanswas varied in diffrnt task was presentedonly one immediately after the sessionsand ounterhalnd ovr sssioS = 0.07. (p pratissSion. 0.s,0.93). Arralvsis o th data idiatd tht there were he intertion of positive and ngativersponss onsiderabl in of difrenes the ets probability in with pobability quite learin thes is data.In addition, In thsc thre eperimnts, xperimnt tlre ovr]l l. tlre negtive R is onsidebly longr tlrn th positiv mn R data slrorvs probbilit efets all (slop no at R. whilr onfirms prviousfindings(Sternbeg, l967. = lI. ffet ws l]). In periment th probabiiit1.' 1969; Nikrson. 1969). h intrti ou l.rvn in nturall tlre sme s in usu] two.l-toie tasks. In data. hlrvever'disagres with tlr results obtained in -perimnt this efetis rL'stord IIl, tlmpletl spit in otlrr binary lassifiation tsks (Strnberg, l969: o the distrto tsk. whilr ws usd again in this Krugr. l970). whr negativ and positiv tsts erimnt. produed th same v]u of probbility efft. h Tli introdution o a distrtor task not onl onsidrablv largerv.lue th probabilitof fetin this inrsd mn RT (sdat for xpeimntsI and III, task an b attibutedto th fat that the tsk involved Fig. 2). but varian as well (see Fig. 4D, th uppr poblemsolving. uves).ln Table 5. rrorsnd omissionsmad by Ss in In ontrol Sssions tlris primnt' o simple R |or solving th distrator poblem r shown. They positiv and ngativnumbrs b1, thmslvs ws hratriz the Ss' lvI f performn on th masued. usd two tyes of instrutions: the W ln distrator task in primnts I and lIl. Sliglrt first. Ss wer rquird to narn th numbr tht had diffens in tlris lvel suggst tirt thr was no ben presnted: the seond. Ss wre rquird to in task in rmarkabl adaptation the S to the distrtor of ..es'. ..n'' answer or aordins th tgory th to Ol xprimnt lII as omard witli primnt I. pesented numbrs.Nithr tlre numberidntiitiorr Thtr.1h onsidrblediffernsin probability time nor tli vrbal resPos tim. vld .ry annot b sribed flt obsrvd thseprimnts in diffrens onntdwith thes two atgoris. h tlr tor. Hypothetially. absene o to this adapttion abov.mentiond dirrr obsrvdnl-rvhn S is is probabilityet primntI ould b eplainedby in problemo deteminig di''.isibility adrvith the th of the ollwing:lr distrtor1ask mkes it impossible numbrs. and tlris diffrn inrssrvhn stim]us for S to orm any adquat subjtiv modl of pobabiI derass. ity p rob abitity struture o stimulus sequn and one way in wlrih th diffrnbtweenth RT or aordingly to onstrut ppropriat ptany and positiv and negtiveresponss binry lssifition in esponse strtegy.However,working during th sssiotts tskslrasben aountedfor is in termsof srialsrh without th distratotask (xprimentII) does nabl mdel (Nikrson. 1969).That modl, howevr,hrdly th S to do that. and th xperingind lrr was aounts for the results in Fig.2D. This qustion, probabl, trnsfrrd perimentIII. to thrfor,rquirssriousthortiland primnt.l study. Neverthless' is tmptingto onjture it that this plrnomenonis due to th ft tht bfor negative (D) Probability ffets with Diffrnt Levls of Motivtion disionsare mde thy ar ..hkd'' on mor,while positivdeisionsaI mad immediatly. Possibly,tlfs is wht uss the inras in tlr diffrnsbtween The influnof difernt levelsof motivationon th positiv nd ngtivRT whn tst numbr probability magnitudeo the probbilityftis illustrtd Figs. in deases, wlren S's doubt about the orretness i.e', ]F to 2I (I-eontjev Krinhik' 1961, 1962' 1964).Th of & negtiv disions may inrease. data in Fig. 2F ar rom an eprimnt using a Figure 2 shows th rsults of a study in whih Ss two.hoitask with lits as stimuli' kyprsses th as performed a two-hoie task in whih lits wr used as responS, and ..equipmnt brakdown'' as th stimuli and kyprsses Irsponses. s Interlavd btwn reinforing vnt. That is. t th bginning of the Suessivtrials on this task was a distrtor task rinformntsssions,th S was told tht if his RTs to

Tl 5 Average Prnt rrors and Avrage Peent omissions in the Performan of a Disttor sk (rihik. dikarov' & Boisova' 1970)


PROBABILITYFFTS IN HoI RATION TI ASKS
limit, th one o th stimLrliedda prdetrmind quipmnthad broken down and that day's sssion was Whn that typ o rinforment was usd, trminatd. unbeknownto tlr S. th swithdoff th quipmnt that pesntedthe stimulus sequen.That type of rinforemntausd a onsidrabldrsin th magnitude of the parametrsthat lrarateizeth probabilityet. Th fat tht th effetofprobabilit signalswas almost the same as in the on nonreinfored wre nonreinford in perinrnt whih both th signals is eden in support of hvpothesiswhih aounts or th deras RT to rinord in signaisin tmsof th rorganiztionand intnsifiationof information prossing wll s in terms of spiJ as mobiliztionof ..inr'' rsours o reativit whih re some developd in rinford sensor-mrtolsystem (Lontjev Krhhik, l962). & that th pfamtrvlusof hs rsultsstrggsted th probabilityfetmit b used to sl tlr legr or his suggstion o subjtive signifin tlr signals. ws onfimed by Lontjv and Krinhik (1964) in an xperiment with two' four' and eit equiprobable wr hoies,in whih thr mthodsof rinformnt used: () inormtion reinforement onsisting of flshing a rd lirt on th S's pnl if his RT to designated sial eeded 400 mse (Fig.2G): (b) shok rinformnt onsisting of delivring an ltri shok in aodanwith the samritrion as usd in () above (Fie. 2H); nd () brakdown rinforment. whih has already ben dsibd (Fig. ]l). orover, rnn ..bakground''sessionsin prededthe three whih all signlswere nonreinord sessions with rinforements.Tlr rsults o ths of periments inditdthat th parmeters probbility in fft ar quit snsitiv to difrns dgr of subjtiv signiian th signals. {nfmation proved to be tli least fetive. ths In reinoment eperimnts. tlr slope dreasdtlniy by 7 .9% as othr ompard to th bakground priment. T1.r produed a sizabl efft: two kinds o rinorement as ompard to tlre bkground prinent. th and by slopesdereasd 28..I%for shok rinformnt As b 21 2% for breakdownreinforement. befor,data from thes primntsinditd tht th ffets of probbility fo the nonrinfored signlsdid not hange signifiantly s ompard with data obtind or th ..bakground'' exprimnt. sign] th in sa.l () Proility ffet and Sequntialffets

139

52o

r
I

.,-.L
f,J

'.-..,,

U)

-"1
L I .-...\. -'..

8l

z
lFiu.

:

4 0l

+ol -.r.
'u/

*i
* l-,/^*\
'rJJl

| j

5oi

-^.r

v-

a

.^I.

vl',

t\ \
i,U


a

20l

0iz3+5078 N UY R 0F ].i.ltNs

1,1,!l

0r't345678

,\*..-

l.^ ?v
o l2 3 + 56 78

L

\
+



o

.,.'..',J

|2

s

78

Fig. 3. a R fo sussive stimulusrepetitionsto stimrr|i whih ourred with p = .5. he tsk wre as follows: anel A. two.digitnumbersto keys (Kihik & Akimova, 1970);Panel B' twodigit numbers to verbl lesponses (rinhik & Akimov. 1970); Pael C, upper htf, twohoie task with distrtors (Krihik' ednikrov, & Borisova, l970; lso se tt)l PelD, lowe half, tones to keys with Donders'Type A, B, ad C tsks(rinhik & Rajvskij,1970;lso stt).

eprimnts tlrat have been disussed thus far re illustratd in Figs. 3 and 4. The data in Fig.3 a al] for stimuli whos probability ws .5;th dt in Fig.4 ar efftswas introdudinto for stimrrli witl-ra probbility o.93. Th dt in the A The analysisof sequntial RT perimnts by Brtlson (l961) and Falmagn panels fo both figues a from a study by Krinhik nd (1965) and beamethe bsis of some impotantmodls Akimova (1970)' in whih th stimrii wer two-digit and or th ffetsof signal numbrs and th rsponss wer button prsses(s ls for reation tim in gene.l Tbls 3 and 4). Th dat in the B panls r from th probability in prtiuiar(Falmgne,1965; Flmagne& hios. 1969; Komblum, l967. l969; lming' |969:' sare study (Iftinhik & Akimova, 1970)' in whih th stimuli wer two-digit numbes and th responses wre Thios& Sith. l970). The eftsf stimu]usreptitionsfor som of th verbl (see also Fig.2C). Th data in the panls of


140

KRINIK

0 LJ ,)

Z
h l:-

|(

. l

'!

-i,o,o

. -Ao.. .{
.,z ? :r*.
\. t.q

NU}.4tl Fig. 4. Men R fo srssiv stimulus retitbns sam studis s th rrsndi peels i F.ig. 3.

oF

R=l|oNs

to stimuli whih (ued with p . .93. Th sks i h pal :e from t}re

Figs. 3 d 4 aIe fom study b1, Kinlrlk alii Alexndrova (l966). in whilt tli stirnli wr gometriftgus (s and th responss wr kyprsss .i}otion also Fig. l). r uppr of th D panis illustratdt om study by Kriilhik, \Jdnikrov. of and Boisova(l970)' in whih th i.tts a distatr t]k wr eltmind in a two.Lloitsk (s also Fig. 2). h lowr potion f the D Panls illustrat dat from the Kinhik and Rjevsj (|9.]0) study, in whih tons wr used as stimuli nd kypressss th responss a DondsA, B, and tsk. in An instion of Fig. 3 mkes the follow.ingrsults (l) quit pprnt. The two.hoieequiprobble data do the strong,monotoni derse R not dmonstrat of with sussiverepetitions, vn in th rang of on tl four repetitins. ]h is in agrementwith prviously reted results (Hyman, l953; Hale, |96.|a, l969). b' (2) Whn th numbr of altrnativs inased is beyond two, th dop in RT with suessive rPetitionsbomes gatrfor the kypressigtasks (Figs. 3A nd 3C), but not for the numbr.namingtask (Fig. B). Th formr onfirms similr findings by Hle (|96,7a). (3) No monotoni derease of RT with sussiverpetitions appears to ou beyond th furth rptitions. In ompaingthe dta for the stour petitionswith th dt of other ivstigtors,similr results I found (Bertelson,l96l; Flmgn'|965; lonrd, Newmn,& l969; arpntr,l966; Hale' |967a,' |969.' I.^aming, 1969 inhik, |969; Ifuinhik & RemiIon' dniov, l970). [n all thse studies,eithr th data we not reported byond four rptitions r the urve approahed asymptot ftr fou rpetitions of th

si]. (a) he distrator task apprsmerly to hv shiftd th ta Jngth intrpt without ftingthe do ll th iuntion. ,he dta in ['ig. 4 ar th man RTs or sussive petitionstl high probabilit sipals = .9). (l) As ws true frig.3. all the dat i Fig.4 (eptor 48) show a lar tndn to bome symptotiftr the o first iour retitios. (2) Th stepnss the urvsin with tlr. r:rgeoi. one t fur petitjons irless inrsing nttmbr a]tlntivs.again with th ptin of the eperimetwith th numbr-naming tsk (4B)' (3) In th erimnts \yith the distrtor gaterthan in task, the RT vriabilityis sigrrifiantly th one without the disttortsk (Fig. aD). (a) In all ineass, the primnts,s the numbr of alternativs R vibiJit dres. ln summary, n nalys of sequential effets or tjghly probabl stimli indits tht sequntialffts r mot omplx thn had hithrto ben thout.

DlsUssIoN
Sin it is impossibl to disussaasptsof th dt within the frmwork of this papr' disussion will b rstited to two qustions that dal dietly with the psyhologial ntur of th probability fft in hoi RT tks. The stqustion dels with th reltionstrip btween the number of alternatives and the relativ frqueny of signalas th two variblsthat dtemine th probbility fft. The sond questin dels with th idntifition of the main dtminants of Th tril-bytrial varibility of R in ths peimnts.


PROBABILITY BFFTS IN CHOI RATION TI TASKS data, prsentd in SetionA (Figs. lA and 1B) prmittd us to sparatth traditional onpt of stimulus unertainty(Hyman, 1953; Briker, l953; Wlford, l960) int two omponents: (l)ltrntiv unertaintyrlated to the number of alterntivs, and (2) tmporal unertainty rlatd to the rlativ frequnyof th signals' Th formrwas interprted a s determinantof S's unrtaintyregarding whih of th alternative signalswould our at givenmomnt;th lattrws interpreted detrminnt S's unrtainty as of as to when,i.e.,at whih point in the stimulussequn' th givn signal would our2 (Krinhik' l963a' b; Lontjev & Krinhik. l964; Krinirik & Aleandrova, Th idntifiation of ths two omponents f stimulusunrtaintyled to a new pproahtoward the undrstanding Hnlan's (l95) results.and to of modifition his onlusions regarding the psyhologialequivln differnt ways o varying (Krinhik & Alndrova,l966: stimulus irrormtion Konblum.l969). he dt pesented Figs.1' 2A. ]B' rrd] provid in strongvidn suort th ntltion that prbbility in ftin hoi RT 1asksis tlr rsultof th joint ation o thes two omponnts of stimulus unrtainty. Furthrmor, adjustmnt of these omponents sets before S's two tasks o dif'frent mplxit (se.t'or istan. Fig. lB). Th ontributionof ah o tlrs oinponents th probabilitffet appars dpend to t lagelyon the mthod usedto .lrystimulusprobability. hus' the rlativelvslit differens th probabilit in l.ts betwn Donders' A nd B tasks(Figs. ]A and lB) wuld imply tht in two-hoi: taskswith unqual stimulusprobbilitis th ftof temporalunertainty is dominnt' In ontrast. vry lagediftrns the i prbability ttbetwen the two.hoie task with nonequiprobable signalsand multiple.hoietskswith (Fig. 1B) impliesthat' in th lattr, quiprobabl signals th t alternative givn of unrtaintypredolnfurats. tht temporal unertinty ws the sme in both situtions. on o th tors towardsulra ontributing signifintdiffrnin th probability fft btwn th two.hoie nd 1h multiple.hoie data is th diffrnin th dgrof prditabiJity a partiulr for signal,hn the dege effiinywith whih Ss an o prosss thse dvlopantiipatoryand prepratory in two tasks.Some observations inditthat, in hoi RT taskswith equiprobabl signals, eitherrfuseto make Ss guesses regrding the net signlor. if they do, it results in too mny errors (Krinhik, 1968; Strnberg,l969; Swnsson& dwrds, l97l). The latter may b du to the dgrdation of th tmporal strutur that aompnis inrasin th numbr o ltrnativs. an It threor seems rsonbltht modls whih dsrib prbability efft in RT would inlud th numbr of altrnativesand th rlativfrequen of the paramtrs. addition,th way thse signals sparat as In would be relatd would depend on the patiula probabilitvwas vrid. methodin whih stirulus

|41

te66).

With rgard to the seond question dealing with the trial-by.trial variabilit of th data, Fig.4 ppars to provid som new viden regarding the behvior of squntialffts in hoie RT tasks. Th absnof numbrof in monotoniderease RT with the inreasing stimulus rpetitions, ven in th ang of th first fiv repetitions, th rpid approah of th uve t symptote, and th strongly pressd varibility of points in th symptoti stion dmonstrtetht this dpndne is th rsult of rathr ompl dterminationtht annot b redued to the simpl as o th influenof th preding signal' of Several onjtures rgrdig the dtmiiants squentil effets in hoi RT tasks r onsiderd below. Thr app t b at last thr aspetsf a sthasti stimulussequne whih ould dtrmineRT on a partiular tri]: ( l ) th immediteiy prding stimuli (extendingftlrn l bak to 4 or 5), (2) some partiulrpttrnso the stimulussqun' and (3.)th f otai stohasti strutuI th stimulussequne. of Th first givesris to the wll-knownrnyets, orrnof a sequential Ild represents most lemntary t]r detrminant humn R priments. in h sond plesentsa mor omplex type of sequntial detmint' whinth S privs sparat stutual pattrns within the stimulus squne s seg]nts nd builds his ptationsin trms of suh sgmnts.Th simplst empl f suh an fft is illustratd biow om dat in a nonquiobabl two.hoi task (Leontjev& Krihik' |962): N.A.. Reord No. 59/l (l.7.196i' Pl = Srrbiet 0.07. P2 = 0.93) ...210. 190, 180, 180. 2eQ. 1-' 60. 240, 40*' 220' 190' 300*' 200, ]90. !q0, r80,180.r70.. Rs to the rar signal ar indiated by biilg rrnderlind.3 strisk(*) denoteslengthnd th Rs t the mor; rqunt signal. This lengthrring may b intepeted bing due to S's petnyof th re s signalat tlris point.+Suh ptterns stimlussqun of, an b imprssdlarl b th S and dtrminhis uthereetations' bauserpetitionsof rat signals wr vry surprising and untypial vnts in th squnes with givenprobbiIity distribution. This fft ould b amplifid lso with th S's rror esponsto the epeated rare signal. These fist two dterminantsmay be thought of as giving rise to short-term sampling of the stimulus squne nd detmining short.trm sltiv prepartion(Brtelson Barzle,l965; Leonard t al' & 1966). h third dterminantonsistsof th totl sequn strutureand its subjtiv reprsntation th form of in subjtiv probability modl (Krinhik, 196a,b' 1968). Th data that would be neededfor a omplete dsriptionof suh a modl are not availablnow, but th model would undoubtedlyinlud; (1) a subjtiv stimat of th differnes betwn th rlativ


t42

KRINIK
Tbl 6 an RT to th Rr (+) and to th Frequent signts.

SignalRu Psition

irst Sssion strt iddl nd 340 226 |9 ,9 19 200

SssionWith Reiforement of requ Signl - .-strt iddl [,nd 310 26 24(l 190 180 190 t83

Sssion With Riforment oi Re Sisn.l stt iddl nd

0 I 2 3 5
*Frqunt

+

290 246 236 2'.t6 253 70 266

310 280 l8 193 20 190 220

300
190 180 180 176 160 1',l 5

330 r76 163
183

r93

116 r56 r66

2 255 266 24t 240 230 245

226 246 26 196 r'7 6 190 20

276 2s6 246 220 206 206
L;

signls o listd ccoriing to thil order lf ouln in runs up to f|v r-pritions (0.i). |ithin ch of th thre sesbs, mon RTs for tiols that follow th|e !ssi|' ocurftns of the rr igna! ar prsentd lor lhe start, mildlc, and nl of sssion. In h ssion th rI |ignal ws prsntd l.S rims nd I|,i. fqut on 225 |ims. Dt o| S N'',|. (Lonrjev &' Krinchik' l92).

frqunies th sigalsl(2) a subjetivstirnto of th proportionof stimulusrptiiinand rlonrptltion, Konblum's studis (1961 i969) dtonstrted ' onvininglvtJt imprlrtan()1. ths vri.bls ih i ,| probabilityf i.; srrlr1etiv t: IPrstatir.)n rh f tmporal strutur (..tim.j rnphloBv.') tht i' essoi3td !vi1hthe oJurn rtiul oi signaland whih distinguiessqunes with diJ|ntstinlulus probb ility distributins (tfuinhik. l969). his dterminant influen.:s some gnral Jevls of t petanyiltid pre'|iiivsttt: th rar nd i tht: frqunt sign.ls,whih hrtriz,eth session as x whol. hs ivls r establish
t.urYsrp!Sent th avrge general lvel of RT to a rqunt tlre signland' otlsequntly. level eativity of the ;iproprit snsry.motr systm(B1,tlko, l964; and Churikova. i967), wlrih is stahlis}rd rgulatd by the S :r th basis of infrmtin bout the tota] sLiuen. sttistiel stluturof th stirnrilus Th l;letil--"-.tril dvitins th iri-lividulRTs liln rhis gralivi rr'pesent f]ututions th the ol. rtiYltyof th givn snsory.mtor s,stm whih ar du to th first tw lteminants merttiondabv. h t1lt dynmisf th S's short-tm Prpratory stt. h irritilstion th urvs.i... th stfive rptitinsof tlr sia!. aB b lgalded s a segmntof a funtiott irr wtrilr Rs e bomig symptoti.This pPers to b nssry,baus at th stIt of this setion,the sensory.motor system,whih is rsponsible ations tc [rqunt signls,is ihibitd by th S's tionto ar sial. It semseasonbIe supps to that tlus stion rprsnts a prorss in whih this inhibitoy stat is ovrom rd the systemrativity is raisd tl an asy'.ptotilevl. This pressan Proed at diffrnt rats, dePnding on spil ftors, suh as the S,s xprinin ths perirnntsor the degreof subjtive sifian f th signls, et. Table 6 presntsdat whih suPport suh n itepttion.As n b sn from Tble 6. th S's RT did not pproah symptoti levl duing th st fiv rpetitions of the fequnt signl at the stt of th stsssion;howvr, to t th nd of tht sssion,it ppas hav don so just aftr th st rptition. [n th sssion whr th frquent sigrl rinford, th asymptoti level apprsto hv bn reahd very rapidly with th st oun of th frqunt sal (se the middl and end of that sssion). In th sssion whre the rar nl einfoed,th RTs srm to pproah th asymptoti levl mor gradually. The onjeturs that aI presented hr in onntion with the natur of the squential ffets in hoi retion time tsks must b rgardd s spulativ and requirig new pimntl onfimation. Nvrthelss, it sms rsonabl to outlin in summr that th probbility efft s wll s th squential ffts i oi RT tasksr the sltof th omplx interation


PROBABILITYFFTS'IN CoI RACTION TIM TAss

|4

t,

L
I

I

dejatelnosti. In D. A. oshnin (d.)' Sistm '' ,,Chlovk.torna,. osow: Nauka, 1965. Pp. |19-12,7. (The rate of gain of infomation and th voluntay regulation Betlson, . Squential edundany and speed i sril ou of humn atity. |n S stc m,,humn. tomt,') two-hoie rspondi tsk. Qualtelly Jounl of Konopkin, o. A. Proisvoloje egulirovanijedjatelnosti po pijomu informatsii v uslovijkh ltrnativnogo vybor. In B. pimental Psyhllgy, |96I, 12' 90-|02' Brtlson, ., & Bazel, I. Intration o tim unItainty and F. Lomov (d.,), Problem izinernoi psikhologii. o|.4" reltiv signal rquny in detmining hoi reation tim. Leningrd: obshestvo Psykhologov, 1966. Pp. l8-34. (Voluntay legulationof human infomtionpoessing th Journ] of primntj Psyhology, 1965' 70' 44845|. irr Bodrov' V. A., Gnkin. A. A', & Zakovskij, G. . Znahenije pwchologv ') n. hoie.onditio |n hoblms lf nnring sttistihskjstrukturi. i odnoznahno-dtelniniovnnot nopkin, o. A., & stjukov, G. A. Verojatnostnoj pogrnozirovanij kk dtrmilanta skorosti rktsij na reagivanii na signal dlrrkh dov. Yoprosy Psykhologii, l966' 6' 77-86. (h signifiant of sttisti.l strutu in alterntivnyjsignaly. Voposy sykhologii, 197|' 3' 46-56. (hobbilisti prognostication as a dterminnt of ratin s o th simply dtrmind sponding to signals of two typs. Qustions of Psyhology.) time to altelntivsignls. Questionsof Psyhology'). Bikr, P. D- lnfomation measurmentand leatin tim. In H. Krblm, s. Choie rtion time for rptitions nd non.reptitions_A rxamination of the inomation Qustler (-d.'),Informtion thor in pry-hologl'. Glno: Fe Prss.1956. hypthesis. |96,1 2"I 1,18-|8.7. Ata Psyhologica, ' ' Brunswik, . Zur ntwiklung de Albedowhrnhmung. Kornblum' S. Squentildeterminants ifmationprossig of Zeitshrift f Psyholoe' 1928' 109' 40.l 15. in sril nd disrte hoi ratin tim. Psholgial Brswik, . Pobability s a dtmie of at beho. Jorrn] Review. 1969.16. 1 I 3-131. o primental syholog, |99 ' 25 ' |.l 5-|9,I. inhik' . P. problemeprerabotkiinformatsiihlov:kom. Brunswik, . orgismi ahievmnt and environmntal In Tzis'v Dokldo II Sjezd bhiestv Pskhologo' g. pobability. Psyhologial Rew, l943' 50' 255-2,72. APN' 193a' 80-86. (on the poblem of humn infomtion poessing.Proedigs of th II All-Union ongess of Byoko. . |' Vrmj rektsii. osow: l{dizina. |964. (Rtion time) Psyholosts. PsyholoJ Siet of USSR' Msow.) huikova, N. l. S/oyo kk fktor uprL:!t1ij v}1sshj inhik, . P. Nkotoryje osobnnosti potsssprabotki rvoj ditlnosti |tIok. sow: Posveshjenije, l967. (Th infomatsiihlovkom.Kandidtskaja dissettsija, GU' . word s ator f th rgultion of human high bain 1963b. (on some asptsof human infmtion plossig. tivity.) Dissttion, osow Stat Univrsity,osow.) Flmagne' J. . Stohastic models for hoi reation time with Kihik' . P. o dterminatsii povdnija verojatnostnoj appliations to experimentl sults. Joun.l f thmatia.l strukturoj situatsii.Voprosy Psyhologii' l968' ' 24.35' (on Psychology. 1965, 12,'l'l -124. determination behao by the pobilistisftuturof tlte of Falmagne, J. ., & Thios, J. An attntion and mmoy in sitution. Questions of Psyhology.) (Als published in ation time expeiments' ln W. G. Koste [d.), Attntion nglish in Sovit Psyhology. vol. 7' No.4' Summe 1969.) d perfmane. II. Ata Psyhologia' 1969' 30. 16-24. Ifuihik' . P' Th probabilit of a signlas deteminntof Feignbeg, I. .' & Tsiskaridz, . A. Verojatnostnoj rtion tim. ln W. G. Koster (d.)' Attntion and prognozirovanij, pednastrojka k distviju i lmja prfoman, At Psyhologia, 69. 0' 21-6, II. |9 dvigatlnoj rktsii. In . N' Kohergin and P. P. Volkov IGinhik' . P.' & Akirnova, . K. Issledovanijffekta ( ds.\, Pro bl m mod l irol' n ij o p w kh ich sko}, d jd t !tlo Sti. verojatnosti sl na vremja reaktsii vybora v uslovijakh Vol. l. Novosibirsk: NGU (Univesit of Novosibirsk)' 1961 . raznoj stepeni sloznjosti vybora i rdihnoj stepni p. 14.l40. (Probbilisti pognostifiation' pepaIlion to sovmestimosti stima i eaktsii. PritsyPy vojatnostnoj respons nd tim of motor retio. |n hoblms of ogizatsii ovedenija. zysy dokladov Vssjuznoj modelling p s ch i c rivit .) onfrentsii)' Lningrad' |9.70' 2.4. (h study of the Fitts, P. ., & Switzer, G. Cognitiv spets of iformation stimulus probability fft in RT epiments with th proessi: I. h familiity of S-R sets and subsets. Journl different degesof hoice and S.R ompatibility. Paper givn of xperimental Psyhology, 19 62' 6' 2| -29. at th All.Uion onfrne on the hobabilisti Piiples of Guln' v. V.' & osnitskij, A. K. subjektivnyj prognos i vrmja Bhaol ogaization, Rognition, nd edil Dignostis. rektsii' |n Motial'' III Vssojuznogo sjzd obshhesto (Thesis' Lningrad') Pskhologov SSSR. Vol. l, t.3. osow' 1968. (Subjtiv Iftinhik' . .' & Alxandova, N' o sootnostrenii vemennoj L. prognosis ad retion tine.Prdings of th III ll-Union i ltemativnoj nopedlnnosti v uslojakh predahi ongrss of Pshologisrs. Psyhologil Soity of USSR, iformatsii chelovkom.Voprosy Psykhologii' |966, 2' 2s -4. Kiv.) (he elationship btwen temprl nd Jtmtive Hal' D. J. Sequntil ffets in a two-hoi seril reation task. unetintis. Questionsof Psyhology.) Iftinhik' . P.' & Alexandrova, L. N. ffekt vrojatnosti i QuIterly Jornl of xperimental Psychology, |96,7a' |9' I 33-141. vemj reshnija elmntamykh ritmtihskikh zdah v Hale' D. J. Sequntil analysis of effets of time unertainty on situatsii binarnoj klassifikatsii objtov. Pitsipy hoi retion time. PeptuI & Motor Skills' l967b' 25' Vrojtnostnoj olgniztsii Povednda' (Tezy Dokladov 285-288. Vsesojuznoj Konfrentsii), I-eningrad, |9'I0' 4-5. Hle, D. J. Reptition and probability ffets in seril hoi (Poability efft and time of the deision of th lmentry atin task, At Psyhologia, |969 ' 29 ' 16-|1 |. arithmtil pblms i binay clssifiation tasks. Pape Hyman' R. Stimulus ifomation as a detemiant of reatio given at the All.Union Conferen on the hoabilisti time. Jounal of xprinental Psyhology, l953' 45' Piiples of Bhao organization, Reognition and dial l 88-196. Dinostis. (hesis) Leningad.) Ivannikov, V. A. Pdnastojka k dejstviju i j svjaz s rinhik' . P.' & dnikarov'P. D. o mekhanismakh vlijnija

. vrojatnostnymprognozom. Kandidatskja Dissrttsija, of svral dtminants and that |he totl stochastic }vlGU,1969. @epration lesponsnd its onnetionwith to Structur of th Stimulus Sequrlc in its subjetiv the probbilisti progosis. Disstation, osow Stat reprsntation takes the prinipal pla btwen them as Univsity'osow' l969.) dtrminant whih must b mst inhrnt in the Konopkin, o. A. Skolost priema ifomatsii helovekom i bhviorof humon Ss in suh eprimnts. sozntIno.proizvolnj reguljatsija hlovheskoj

RFRNs


|44

KRINIK
flekskh.Dokldy A.N.sssR, 1935' l' |9-200. (on the possibility f ontat betwrn ti probability thory and the thory of ad. IJ. Pvlov on onditioning. Reports of the Acadmy of Sinsof th USSR.) .h Skinne, B. F. bhvior of orgnisms" New York: Appleton{entry{ofts, l 938. Sokolv, . N. Vosprijatije i uslovnyj reflks. smw' GU' 1968. @erptiond onditioning.osow State Univrsity, sow.) Sternbrg,S. Two operations in hratr rognition: Sm evidne from rtion-tim mesurements.Peptio & Psyhophysis, |967' 2, 45 -5. Stenberg, S. The disovry of proessing stges:trnsions of Donds' mthod. ln W. G. Koster (d.)' Attention d peformane, At Psyhologi, II. l969' 30' 276-15. Svadosh,A. ., Shndik'., & Jampolsky,L. T. Anliz odoj modli adptimogo povednija. ln A. N' Kohergin and P. P. (ds.)' hoblm modlitovnij pikhichskoi Volkov NGU' l968. Pp. |68-I.|2. Vol.2. Novosibisk: dejtlttostL (The Iysis of som model of adaptive behavior. In hoblms of th modeiig of pshic rivit.\ Swensson, R. G.' & dwrds, W. Reryonse stlategis i two.hoi atin task with a ontinuous ost for tim. Journal of pimentalPsyhology' 1971' 88' 67.8l' in Theios' J., & Smith' . G. SequentiJdepndenis hoi reation tims. Rort No. 70.8' Wisonsi Mathemtial Psyhologyham, Univsityof Wisonsin'1970. hma.s,. A. on xpetany nd avIageation tirn. British Jrnaloi syhology, |970, 6|' -8. \Ylford, A. . The masuimentof snsory.motorprforman: of Su.ey and rappraisl twlve yars'ploglss.rgonomis. l90' 3. |89.229.

verojatnosti signla n \rmja raktsii hlovka. Voposy Psykhologii' !9,Io' 6, 3446. (on the mhanismsof sJ pobability fft in oie retion time experiments' Questionsof Psyhology.) Krihik' . P.' dnikrov, . D., & Boisov, . . Detemintsijpovednija hlovka velojatnostnoj stuktuoj situtsii v uslojakh djstvija ..pomkh.'' instipy Vrojatnostnoj orgnizatsii Povedenija (eris Dokladov Vssojuzoj Konfntsii).Leningrad' |97a' 37-38. (on th dtmintion of human behavior y the probabilisti strutu of th sitution in the tsk with th distto.Pape givn at the All-Union Confrene on the Pobabilisti Pinipls of Behor orgnization, Rognition and dial Dignostis.(hsis) Lningrad.) Kinhik' . P.' & Rjevskij, A. . Zvisinostvremni reaktsii ot ..obnaruzjenija''' verojatnostipredjvlenija nal v r-rslovijakh ..raslihnija''i ..vybor'' signl. Pritsipy Verojtnostnoj oganizatsii Povdenij (zisy Dokladov Vssojusnoj Konfrentsii). Leningad, l970' 8.39. (Probability efft in Donders' A, B and C rtion tsks. Par given at the Jl.Union Confen on the Probabilisti hinipls of Bhavior organization, Rmgnition nd edial Dinostis. (hsis) Leningld.) i inhik' . P.' & Vilenskaja,. G. Reaktsija..ozjidija,' eje elktromiografihskoi vyrzjnije v uslovijakh verojatnostnoj id' l970' 35.37' organizatsii stem signalov. (Ptn d its ltromyogaphialrepresntatioin th R exeriments with the robabilisti stutur of th stimul sequene') rugr, L. fft of stiulus proability on two.oie ration time. Jounal of perimentlPshology'1970' 84' 317-,l9. I.ming' D. R. J. Subjectil probability i hoie ration tim expeiments. Journl of athmatialPschlogy' 1969' 5l' 8l-100. Lonard' J. A., Newm' R. .' & Cpentr, on the hndling A. f hevy bis i a self.paed tsk. Quarterly Journl of primntlPsyhology,l966' 18' l30-14l. Iontjev, A. N.' & Iftinhik' . P. o primennii torii inlmatsfi v konkretnopsyhologiheskh issledovanijakh(sovmnnye issldovanija reaktsii vybora). Voprosy syhologii, l96l' 5' 2546. (on the appliatioof infomation theory t orete psyhologil investigations.Qustions of Psyhology.) Iontjv, A. N.' & Kihik' . . o nkotorykh osobennostjkh rotsss perrabotki infomtsii hlovkom. Voprosy Psikhologii' |962, 6' 14.25. (on sm aspets of huma infmation plossig. Questios of Psyhology.) (Also publhd in nglish in Soviet Psyhology & Psyhitry' 1963' |'24.0.) I,ontjv, A. N.' & rinhik' . P. Nekotoyje osobnnosti protsessa pelbotki infomatsii hlovkom. ln A. I. Berg (d.), Kibmtic, ms|tlnije, ziizn. o*ow: ysl' l94. Pp. 221-24|. (Some aspets human infomtionproessi. of |n Cbernetics, thinbing lif.) Moway, G. H. Choi lation times fol skilled rsponses. Qrrterly Journal of xperimentl Psyhology, |960, |2, |9.202. Niksn, R. S. Binry-lssifition respons tim, mmory srh nd question of sril vrsus parallel prossing.Papr psntd at 19th IntmationI Coress of Psyhology' lndon' 1969. Rmington, R. J. Analysis of squntil ffets i hoie ration tim. Iounal of perimrrtal Psyhology' |969,82' 250-?57. Romnov, N. A. o vozmozjnosti kontakt mzjdu teorijej verojatnosti i uhenijn akad. I. P. Pavlova of uslovnyh

NoTs
of 1. he abseneof this efft in the quiplobablesessions th B-reatin tsk n b aountd fo by the well.kown phnomenon of an inrs in the snsitivity of th anlyzers ouling whn n indiffeent stimulus (requiring no ativ rspons' as it was with on of two signalsin th C-ration tsk) bemes signl stimrrlus (requiring n ativ lspons' s it ws with this stimulus in the B-ration tsk) (Sokolov' 1958). 2. It sems useful to distinguh th type of temporal unetainty from th taditionl onept of tim unertinty whih dots th v.ibility of intstimulus intrval nd foreperiod. 3. Th sond ourneof th are sal resulted i an eor (_). (RTs wer measudto tlte narst .01 se). 4' Th quantitativ anlysis of these lngthened RTs to the fruent signls indiatd (Kinhik' l963b) that they wr not th result of radom flutuations. Furthrmoe' the. Gs reorded in the ous of a trvo-hoi task with nonequiprobale signls (Krinhik' l 968 ; inhik & Vilnskaja, 1970) indited tht these lengthned RTs wee ofte aompanied y ineased toni tension in th musls Isponsil for the ration to the rr signal. Ths Gs developd during th intstimrrlusintrval whih preeded the tial in whih th frequnt signl oud. @eeivdfo publiatinJnuary 22'|913; August 27' 1973.) rvisionived