bee er carer site ha me) aia? oy ied niet 7 Sed irik wade at tiene Perea Mi et ti" i Soe ¢ 4 pate Pir Wa: ails F anal aeat} we tgice 8 Sea rans te ef ese jee anled Oy pe aes $5 j ; ita hea rich , “ boat? hha) eA Aaae ne hey US feted tal tnt i Sabi ay By eee er eT Pi bat eand if erat yt wt “4 leaded ‘y mit ven nity + pete Hei ri eit } We Fee Pat tat we Bat 4 ir het + HEE tepveate tt ie) TET eR Dy Bet, eS header mopar Tae H GS 4p re unis doing BAM a fs if ‘ eel he pei ey Aes H ofan be wit hb ett BB Sehvb v9 ha by * Ae why ’ db yas lie rs) ey) He Meads a ea SU Sip psa Pepe east nee Dita tasieat eee vig PVPs penne oe 5 rate arent Hab lhe esitrse sts “ v5 = greeted Wh tarane via) ? $ nip? Ma Tyke Pn ah an tla ty Foe Hy Habseegrss hota doimge Wap)? Coie cag home Saas Fb ayes soe etches) ” anaes ont Cae ES Pa witb dt hte geile Ohya, HPV p dak te nahin By pariarh rh) i ree buibda Thal ge elk b va By yada te haya v 10 he ie at eb Prete erie Pages Wee sh bear be oh i rif fie j se ait ; pnd fines at ¥ spear Lh ed sky rep Lo} id 1Pat> yh ais et th ey Metter a qniee site ; i fl Tee Lad te at iene Pity Bian ait Loewe + ieernh es H} ie at eatetie. Terie Sen Hien tis Hala sianee <3 ct 25 eo eate = ratty fal eh 6: 7 Fi Pe A ey t hs pate y ¥ .u e4 they) dabon riya tt! f\ ef at koi) 4 Se Ie 1M Spies SSeS fe (a atas=5 i — ans aon rh 1G tity ’ RC ry} ae ify 4 ut i pater pr eae ad A Sere motetaes Taree Se: seater: a i 7 aro ete wae ‘es a Stele = noe Beret Une! “4 fat eet ees eae =e ‘4 i ea i oy PR ‘ ; i Pt EULA ach Rae », nh ANG i 4 re Nb i gs phe a had Or arte earnetate Tahoe ih ae 5 tah BHAT, pricene aeey dee epee te APR by a Be a2) cay eet Pa tAtistentt ad Sole t ULE a ; paae ties ye tabs) OBC steal ePrice hg ub Pa pateie ah iF Heal iE ith ety sae ae here HED E x asta Aarne fil at ee ih Preis ht ever dest yale bee eat ‘ i M Wek ff alas NT aes eR SARA DEH ee tet if Ht) ab Le nutes ra ve erty ‘ ry i AAPL EARS, sbadeas, tH evades iM ‘iG 1s 14 sh, / 1! Maiti ae f aya daveaine esti es Th : eat a % $3 be Part} gates ANeatatet * cee dire ini HOPy rat ku wee ainda " Initiate ee iS fie i is 9 0 pet Abe (hy sae oT Bia Heat \ + i ¢ 3 + tings ith Matra yogeine dale basen ine hh He 14 2s a % shat Mianeelds 4 Miikats see nia dey agcaey athalty Pir aye Doren ais beyaley ry ttn Nyon aeut ib ai i) tietasan neers Stet paragon aii 7 Spahanetd ee pia dey be bal Ehahat radars in yi tet iy libakaper treet sees wit LU best) kur ney My ia una ath eRe nf pegged te ; ¥ a AST 4 \ ey vi eee! ey fly cy eis ian) oy) vere op ae ese At rr oh AVEAY r A he he ROM Foe ey ACA Ve ele Sh det rN Pe Fat: BR a UG Ai mL Sa pus wt . i hs r . ph 2 at ‘ ff Ww NOSES iia PSYCHOLOGICAL REVIEW PUBLICATIONS hed ne 16l Psychological Monographs EDITED BY SHEPHERD I. FRANZ, Univ. or Catir., So. Br. HOWARD C. WARREN, Princeton University (Review) JOHN B. WATSON, New York (J. of Exp. Psych.) MADISON BENTLEY, University or Itiinots (Index), and S. W. FERNBERGER, University oF PENNSYLVANIA (Bulletin) Further Studies in Retroactive Inhibition Sa, y 7 ERNEST BURTON SKAGGS, Pu.D. College of City of Detroit PUBLISHED FOR Tue AMERICAN PsyCHOLOGICAL ASSOCIATION By THE PSYCHOLOGICAL REVIEW COMPANY PRINGELON GN, J: AaAnp ALBANY, N. Y. Acents: G. E. STECHERT & CO., Lonpon (2 Star Yard, Carey St., W. C.) Lei1pzic (Hospital St., 10); Paris (76, rue de Rennes) PREFATORY NOTE The experimental investigations reported in this paper were begun in the fall of 1920, although interest in the problem and some preliminary probing (largely by group methods) goes back several years previous to that date. This section of the work was completed in the summer of 1922, being carried on continuously during the interval. During the progress of these experiments several monographs and articles appeared dealing with certain phases of the work reported in this paper. However, it seemed worth while to continue the work as planned, believing that addi- tional experimental data would either help confirm or refute present theories and experimental conclusions. In view of the fact that several recent investigators have given very complete reviews of the past theoretical and experimental work done under the caption of retroactive inhibition (see bibliog- raphy at the end of the paper, numbers 1, 4, and 5), we will omit this customary procedure. However, we may note that many specific problems’ yet remain to be investigated under the heading of retroaction. For adequate treatment of these various prob- lems far more work remains to be done than has yet been reported. We wish to expressly state our obligations to Professor F. C. Dockeray, head of the department of psychology at Ohio Wes- leyan, who not only served faithfully as a subject for two years but did all in his power to see that adequate apparatus was pro- vided for the work. -Also we wish to thank the administrative officers of Ohio Wesleyan whose kind and generous regard for research made the present work possible. To Professor Pills- bury, under whose direction the work was done, we wish to express our appreciation for his various suggestions and encour- agement in the work. Last of all, we must express our appre- ciation to Mrs. E. B. Skaggs, who not only served throughout as a trained subject but also acted as experimenter when the writer worked as subject. ili 1 4 ih, ea ey is a Nal TREE NN SRS Bed Ue WO CP ae eee 5 ' , } Digitized by the Internet Archive in 2022 with funding from Princeton Theological Seminary Library — https //archive.org/details/furtherstudiesin00skag _ ; ‘ TABLE ORS CONTENTS PART PAGE PRE RA TORVENOTE) foe oc thats re eke ee tt eee ee ea ea ce ili I. INTRODUCTORY STATEMENTS, DEFINITIONS AND As- SPPAA ELTON See sis 12’ Cavan sh Arcane eae aR Rue ge he 1 eMC GUN STRUCTION “LEST. Mie ee men etme oo uare ice > A. Comparison of Work and Rest Intervals......... 5 B. Retroactive Effects under Conditions of Fatigue... 14 Crh feciiof Practice on Ketrodcttony.. Ma. ak eo os 18 D. Effect of Varying the Temporal Interpolation.... 20 E. Degree of Similarity Between Learning and Work 25 F. Qualitative; Nature of the Rest and Work EULENVIOUSE Oot eae eee ae Ra ets, eae ee a2 TU etna EN Sia VY RUSS 2 ocr acl eed Mayas MM Belle Raat ere ni eve leds 35 A. Comparison of Work and Rest Intervals......... 35 dead ON etal ig ay NEG ght at Yor Aponie n OP Da ene Cs Sah At A 39 C. Postiton of Expervment in Day's Sertes......... 5) D. Effect of Temporal Interpolation............... 40 PREELON GENS OV LT ABLES 0d yaa, bale daig, Mio) « 4 als sbepukdiats 42 A. Comparison of Work and Rest Intervals, Single NETS A a Cope ee eek ehipert cs Co at + Sa LER a 42 B. Paired Associates, Rest and Work Intervals; Sim- arity of Work and Learning. «.....6....6.6) 45 C. Single Syllables, Similar versus Dissimilar Work.. 49 D. Effect of Varying the Temporal Position of Work, POEIEG IA aS VILODLES 9 a le aanate Cine abd ido Sees 50 E. Effect of Varying the Temporal Position of Work, BaredaAssociates: Method ncisiaee Ca neta ey 52 V. GENERAL SUMMARY OF INVESTIGATION............ 56 Te BUIOCE ABE Vernal, is Goby inl, Aa ome ine anna Caan, 59 we Ni i ¢ jah te aly a * ay ‘ Vp sth rent ef a a ry tia aN " Wigh tay a j i+ te ite yi yt 7 bs ihe Wie 4 PART I. EXPERIMENTAL SECTION INTRODUCTORY STATEMENTS, DEFINITIONS AND ASSUMPTIONS Meaning of the term “retroactive inibition.” In this paper the terms retroactive inhibition, retroactive interference or block- ing, and retroaction will be used interchangeably. If any given mental (neural) activity B, following a previous learning process A, works detrimentally upon the retention and recall of learning A, we denote the fact by saying that there has been retroactive inhibition. We are using the term, not as any theory, but merely as a handy grouping for a number of phenomena. In psychi- atrical work cases are reported under the heading of retrograde amnesia. Here some physical shock or emotional disturbance seems to blot out the possibility of recall of events just preceding the shock. While these cases may involve very different prin- ciples of explanation from those facts reported in this paper, we are inclined to classify them under the heading of retroactive inhibition. On the more normal side, if one forms a given series of associations and then turns to other vigorous mental work (with or without emotional aspects), and finds that the original learning is recalled in an unsatisfactory way or not at all, then we indicate this fact as due to retroactive interference. Subjects used in the experiments. We may classify our sub- jects into three groups. One group may be called the “ trained group.” F. C. Dockeray, Mrs. E. B. Skaggs, and the writer constitute this group. All were trained in the art of giving keen and thorough introspections and were able to adjust themselves to the conditions of the experiment. The second group may be called the “semi-trained group.” They were students who were taking advanced courses in psy- chology following a year of general psychology, involving labo- ratory work. One of these was a graduate student (R. M. B.). Ali were faithful in trying to live up to the conditions of the ‘ 2 ERNEST BURTON SKAGGS experiments. Their introspections were usually inferior to those of the trained group. The third group consists in untrained subjects taken from the writer’s classes. The number of records from each of these subjects was usually small, and hence we must rely upon the central tendency of each group for indication of the actual facts. In all cases these subjects were used as checks, although their results were often quite convincing. The work and rest intervals. As we must in some way com- pare results under conditions where mental activity follows the original learning with those results where a period of rest fol- lowed the learning, it is very necessary to control these two mental conditions. In one case we want an extreme of mental struggle and attention. In the other case we want the subject to be just as passive mentally as possible—to relapse into a condition of lazy, passive perception. We have used the term Work Interval to denote a period of vigorous mental activity. The term Rest | Interval denotes a period of relative mental quietude and relaxa- tion. We have made no attempt to secure a graded series of mental conditions ranging from one of these extremes to the other. Our best rest period never represents complete passivity (this presumably would be death), but merely a condition of relatively little thought and effort. We believe that, on the whole, our subjects have been fairly successful in taking these two mental attitudes, one of work and one of rest. Each subject was told as clearly as possible our conception of a good rest and a good work period. In the rest interval the subject was urged to relax and assume a lazy, indif- ferent attitude. He was told not to worry about the score he made and to get away from the learning material as far as pos- sible. If at any time some part of the original learning came to mind he was to take a discouraging attitude toward it and get it out of his mind as best he could. In the case of the work interval he was told to work as if his “ life depended upon it.” Introspections. The introspections are absolutely essential in the following experimental work. Without these introspections the experimenter is unable to classify the rest and work interval. FURTHER STUDIES IN RETROACTIVE INHIBITION 3 Only a complete history of what went on in the subject’s mind can permit the experimenter to say that “this is a good rest period free from any return to consciousness of the original learn- ing material,’ or “this is a fair rest period in which fragments of the original learning came to mind,” etc. It is upon the intro- spections of the three trained subjects that we have relied for the most part. Our group two (semi-trained subjects), however, contributed valuable introspective reports. While the introspective reports have not been given in this paper they have been taken carefully and in detail. Several tables and important conclusions have been based entirely upon the introspective reports. The use of short intervals. Our experiments everywhere employ very short rest and work intervals. Longer intervals were used at first (e.g., fifteen minutes) but very quickly abandoned. Long intervals bored our subjects and made it very hard to con- trol the rest interval. Long intervals put a premium upon all sorts of mental activity. Again, it was almost impossible to get the detailed introspections which we desired after a long interval. Program of. investigation. All experiments were performed individually. The writer was experimenter in all cases excepting when he himself served as subject. In the latter case I. D. S. served as experimenter. The so-called “ reconstruction test ”’ was used as test material in Part II (following). In Part III a little work is reported in which unconnected sense words were used as learning material. Nonsense syllables constituted the learning material in Part III of this report. We have attempted to gain some further data upon several major problems. First, Is there any definite evidence of retro- active inhibition? Here we have compared results after work and rest intervals. Second, Does the temporal position of the interpolated work affect the degree of retroaction? Here we have varied the temporal position of the work interval while keep- ing other conditions as constant as possible. Third, Is there greater retroaction when the work material is similar or dissimilar to that of the original learning? Is there any relation between 4 ERNEST BURTON SKAGGS the degree of this similarity and the amount of retroactive inhibi- tion? Fourth, Is retroaction greater in the morning or eve- ning—when the subject is relatively fresh or when fatigued? Fifth, What is the effect of practice in a given learning material upon its susceptibility to retroaction ? PART II. EXPERIMENTAL SECTION RECONSTRUCTION TEST AS LEARNING MATERIAL Experiment Series A. Comparison of Rest versus Work Intervals In this series of experiments our problem may be stated as follows: Is there any positive evidence of retroaction under the conditions of our experiment? These conditions are: using the reconstruction test for original learning; short exposure (learn- ing) of material; short interval of work or rest; addition as the interpolated work; and using the central tendency as index for comparative purposes. The reconstruction test was used extensively because (1) other workers had used it to some extent (DeCamp and Robinson) and (2) because we found that it lent itself admirably to the experi- mental needs. Apparatus and Method: A chess board 12x12 inches was framed and covered with glass. Five chess men were used, viz., the queen, castle, bishop, knight, and pawn. A large white card- board 20 x 24 inches was arranged to slide up and down on a horizontal rod above the table, being used to screen the chess board and experimenter from the subject when needed. A stop watch was used in the timing. The subject sat on one side of the table, the experimenter on the other. The screen being down, E. arranged the chess men on various spots, under a prearranged schedule. At a warning signal of “ready” E. lifted up the screen and exposed the chess formation. S. was then given fifteen seconds in which to study the arrangement. The card was then dropped. The experiment now took one of two forms, as the case might be. Either (1) S. relaxed and was at rest for a given interval or (2) S. worked attentively adding columns of two place numbers. Three differ- ing intervals have been used, namely, a half minute, a minute, and 6 ERNEST BURTON SKAGGS a two-minute interval. At the end of the given interval, work or rest, the screen was raised and S. tried to reconstruct the forma- tion. The time was recorded from the time S. began to recon- struct until he signified that he had finished. S. added aloud and a record of his performance was kept. However, the amount — of work thus done in the work interval is nowhere given in this paper, as it was found to be a relatively constant factor. The errors are calculated in the same manner as that noted in DeCamp’s work. Thus, if the queen was located by E. as four rows up and six columns over to the right our original record read simply, 4-6. If S. replaces the queen as 5-7, then the total errors are one for the row displacement and one for the column displace- ment, or a total of two errors. On any given experiment day six individual experiments were done, three of work and three of rest. The intervals were alter- nated for the following subjects: F.C.D., I.D.S., E.B.S., M.S., F.Bl., and Deus. Consequently each subject in this group knew what experiment was coming next. In the case of subjects © R.M.B., R.S., W.A.D., and Sh., the experiments were presented in irregular order, and hence these subjects did not know what to expect next. An interval of from two to three minutes elapsed between each individual experiment in the day’s series. The following tables (I to V, inclusive) give the Means and M.V. for the ten trained subjects and for ten untrained subjects. (It is to be mentioned that only subjects I.D.S. and F.C.D., and E.B.S. knew definitely the nature of the problem.) For the most part the averages for the trained subjects are based upon twenty experiments for each work and rest interval. As we desired some data of a comparative nature between morning and evening conditions, three subjects have thus data for both these times. 7 FURTHER STUDIES IN RETROACTIVE INHIBITION *yafqns yore 10; owes jay our, ‘“Szoofqns Joyo IO} ZI pue OT UseMjoqG Sqoa pure “S'a'l “S'H'A 10} “H'V 6 pue £ UseAM}0q GUOP syusullIsdxy ‘sjuamiiedxe 6 “YS 10} pue ‘sjuewttedxe gl “S'y 104 ‘sjuswiiedxe g “QVM JOF. tsyueutiedxe ZT “q' WA 40} fsyuouttiedxea ¢] uodn paseq ‘snoq 10} soseioAy ‘syuowliedxe QZ uodn paseq st ‘[q’y pue SOW “SAL “GOA “SAA OF wosrs asdeisae YoRY—'SaALON AYOLVNVIdXY ZZ es'¢ l71 LOZ 19° cpa et 88°9 Zb'6 S6'S +0°6 IZ’ cca 16°19 18'r8 +S'°6S 8£°06 S91 Zve Pte. Siete Fee 559 Pe i Re ee Ree Ul pat ote Bree aby Oe GPa. anederitwstda un sib waidunt ted os eye yes ane Neer i reper te eed ete eee ene eee ic Pee eee! es Bre OT ceseensttereccatpeneseetce OL ne nee a reer Sa es eee cea ree eer ee oe ee gee aes te i ae Shi Rte Re 2 pete-sapees smoccr toner simmey 06S 088 ss 0F6 ge 0s°9 OPS O0zOr 062 O00 06° FS oos = ors = 0g'9 Ce SOU = tees 20rS og9 OZIT O0%f Ob pes eae ozs Ors e7 0st Oo OS'Y oes Oror Ooze oer ice Sie ly9 0L£8 6S Osst oF 00% COO S06 eh SES Sab Sarees eye wae CATE Se MAW Se CA = ae CAN - aS A ee SOY FIOM SOY YIOM ysoIy YIOM JeAToUy 0'Z JeAJaqUT ,0'T JeAroqUT ,¢°0 GHaAGISNO)) SUOWWY AINQ ‘“SddOoaYy ONINYOJWT “STVAYWHIN]T LSA ‘SA MYOAA OINIYVIWOD) “LSAT NOWONYLSNOOAY ‘sLoafanS daNIvay, Qf wood SNOILVIYVA NVdJY INV I HIdVL AUVWWNAS TC CY 7 sere e eee ROT aden mete SeH bi Ard MUG Gy cy ed a xy ya fqng SHOVUIAY ONIMOHS ERNEST BURTON SKAGGS "I a[qeJ, 10} sajoU VaG—'SALON' AMOLVNVTAXY 02°02 7192 €S°ZT S861 9¢'8 g60L CAN CcO'ZS ‘ 10 rZ cogs 10°99 S6'SS 99°S9 Se ee ueay{ cers 1999 o$SS 1099 8°E7Z 69ers ee Sl nrelemmiewsceie olan © poi feeje"'s S05 dive Wie, os 691 OS Zr Of'ZZ 0z'S9 Pat his os RSet aby sO eh alot 5 ple Soma FER Nawees ON eT eS 09°IZ os'e8 Of Te 09°86 lvl SZ"b8 Ores 09°S6 ele dine -0" Siufone® o6).0 tals), GES 1slahS ie Te BP Ss OF aa etn ase 08'Z Ob'SZ 09'Z Orez 9°¢ OV'Zz 00°9 Os Te Shs ve \s\'6 6 stops esien cas USan, Wee Os eye S..6 eee tO ACE Rg be Yt 0Z'FI 0S'6r 09°Z1 06'°0S $'6Z 06'8S OF'6I 09'bP Wrosetehene cae Ws eS SP MSE Es SE ce rs fa ios 00° I+ 0S°6Z OS'S 0°66 26 Oe'g¢ oles 0619 RG aon ee ce Sie te See A ee Ne Tg i 06'¢ 0S'2Z 0S'6I OZ'ss 08 Or'9z Of'ST Oz'9¢ Eso SRLS CT 6 Te eG MON mee Line ee a Mel STATE Ey 00ST O0T8 00°F OOETT O'ce 0076 00h OOZIT OO8E O00ZZ 00°98: = 0018 so SS ee SL Of Le 009 09°0r 09°96 Tol og Ps OO'0g 0292 00cc 8 OLS DG or Ue VL Soe as eee Se oe O96 O08'6F ogze 09°08 Sor Of Or Oc re 09°95 OL Sie 0C.se On See LC pe ene Gat OV'8z 05°89 O9T? 0002 Gee S619 0292 002 OV6l 0226S Oy oe a OD raat Se et 9 ae Fi) Sees D AW il, aN Soma Fi AW AL 58 |p eee AW nf Soy YIOM SOY YIOM SOY YIOM JeAtoqyuyT {07 [eAtoJUT {OT [eAtojuy ,¢'O GdYAGISNO) SAWIT, AINOQ ‘SGNOOTY ONINNOWL ‘“STVAYTLIN] 1say ‘SA MYOAA ONIGVdWOD) “ISAT NOMLONALSNOOAY ‘“SLIAf@NG GANIVAL Q[ AOA SNOILVINVA NVaJ{ GNV SdDVaIAY INIMOHS Il ATaVL 9 FURTHER STUDIES IN RETROACTIVE INHIBITION ‘TIT 2192], 10} se oureGg—'saLON AYOLYNVIdXY Oem OF OF = OU PS OT ZS 0S e028 Ube OR Oo = ce. 00S = 0CIh tecL = eT OOS 00'S - — 006Z_ -00'S9 = -00ST. 00:9 08°02 0 Sr == UOT “O0E ~ “009 0008 2°" = = doa O8'6I 0529 O02 Ob OLEE OZH9 OZHZ S99S O8IZ SS09 O8'8I Sees “CCS aa Ne AN eas AL ee AA AW 2b soy YIOM Soy YIOM {SOY YIOM [eAIOIUT ,0'7 JRAIUT ,0'T [eATOIUT ,6°0 dauadISNO7?) SHWIT AINQ ‘SdXOORY ININAAY ‘STVAYALNT LSay ‘SA MYOM ONTAVdWOD “ISAT NOMLONULSNOOANY ‘s~oafang GaNIvay, Q[ Wd SNOILVIYVA NvajT GNV SdDVaEAY DINIMOHS AI ATaVL ‘W'd 6 pue / UdaMJoq BUOP sores yUOUTIOdxe [PY ‘arqey a]OYM 94} IO} SyusUTTIedxa QZT JO [e}0} B JO [eAIOJUT 4SAI pUe YIOM YORI 10J szUoWTIedxe Gz uodn paseq saBeiaae [[Y—'SALON AYOLVNVTdXy oe oe Or9 0£6 66€ Of oc9 O08Z O81 S6T OCS ee ee eee Sl Orr 09'€ 00Z O90 O£2 062 C0 06D ar (b= 026 00Gr SS6Ue 2 eee 09r 0z~Z 089 SrZzl 0%b SLPr ors O82 OLb OFS 00'S eee pee ene OE Na Od eee A Nee Sa SAW Ie MN: 2 ae NN “AW 3a {soy YIOM SOY YIOM {SOY YOM, Jeasoquy ,0'Z JeAJaqUy ,0'T jeasaquy ,¢'0 AYAAGISNOD).SHOUNY AING ‘SNOOTY ONINAAW ‘STVANALN] Lsay ‘SA MYOM ONTIVdNOD “LSA], NOIMLONALSNODaY ‘SLoalaNG GANIVAL, QT AOA SNOILVINVA NVAJY ONV Sd9VaIAY DNIMOHS Ill HIdVL 10 ERNEST BURTON SKAGGS TABLE V SHOWING CoMPARATIVE ReEsuLTS AFTER WorK AND Rest PErtops For 10 Sex Female >. v2). Remaleyeric ae. Mailer, see scaaiebom ates Malewectann Females sce Male... Female. . Female. . otal M UNTRAINED SUBJECTS. AVERAGES BASED oN 6 RECORDS FOR EACH METHOD PER SUBJECT. 1’ INTERVAL USED Work Interval Rest Interval Br ye Tine Oy Er. otime @) 8.5 48.8 Zeal 24 1152 41.2 10.0 36.2 10.0 143.8 6.8 125 4.5 88.7 5.0 87 9.2 103.3 3.8 93.3 F a By AMELIE ERS aide 12.0 5755 8.5 5132 Femaleniy (00/5 seu anies siete erences tas Ons 44.0 6.5 67.0 9.3 7a adh S/o Uh vehi tonts atten ike 10.8 27.0 3.8 24.7 Malencnde rs) 38.8 or3 SSIs. 90.7 663.0 58.5 578.8 APE EWS seg wpa tei Gena ae ls 9.07 66.3 5.85 57.9 see teat 1.61 28.7 Ze3r 28.2 A study of the foregoing data indicate the following: (1) As regards errors made, the case is quite clear for all intervals used— in the case of trained subjects. The performance after the rest — is much better than after the work interval. (A) For the half-minute interval the mean error is more than (B) (C) twice as large after the work than after the rest interval (8.55 as compared with 4.21). For the minute interval we also find the average error clearly larger for the work interval (9.04 as compared with 5.95). For the two-minute interval we likewise find more errors after the work. (2) Considering the above data with regard to individual per- formance rather than group average, we find: (A) For the half-minute interval all subjects do markedly (B) better after the rest interval. For the minute interval, where we have ten cases, eight subjects do better after the rest interval. Deus., one of the exceptions, persistently reported an inability to take a restful, passive attitude, and we suspect that her so- called rest period was as strenuous as her work period. R.M.B., the other exception, reported a similar con- FURTHER STUDIES IN RETROACTIVE INHIBITION 11 dition, with the added fact that the adding was very easy for him. ; (C) For the two-minute interval (nine subjects) we again find two apparent exceptions, namely, Deus. and W.A.D. We believe that we can easily explain the record of Deus. (as above stated). W.A.D. has a mean based only on eight records and moreover could hardly be called a trained subject. (3) As regards reconstruction times, our summary table II shows that the mean time is consistently longer for the work than for the rest periods. (4) The evening records for subjects F.C.D., I.D.S., and F.C.D. (Tables III and IV) show the same thing—there are markedly more errors after the work than after the rest interval. (5) Table V gives the results for ten untrained subjects. Here only the one-minute interval was used and the records are based upon only six experiments for each work and rest interval. Inspecting the group averages, we find (1) more errors are made after the work than after the rest intervals (9.07 as compared to 5.85) and (2) the reconstruction time is longer after the work. Looking at the data from the standpoint of individual performance, we find that eight out of the ten subjects do better after the rest interval. Supplementary Control Series (A) It may be objected that the reason our subjects did more poorly after the work series was due to the fact that they were fatigued. Or, again, it may be objected that directly after the work the subject was so far removed from a “ reconstruction attitude ”’ that he could not get back into this attitude easily and so gave up too quickly. Our work intervals were so short (two minutes being the longest) that the fatigue objection seems out of the question. As six experiments were done in succession, with a small interval between each, there would have to be a most 12 ERNEST BURTON SKAGGS remarkable oscillation between fatigue and recovery for these short intervals—something quite foreign to what we know of nerve and muscle fatigue. Muller and Pilzecker and Heine have also demonstrated that the above objections are largely insignifi- cant. As regards the attitude objection, we have often found that the subject was “ farther away” from the experiment at the end of the rest than after work. However, a control series of experiments was done, largely to meet the above objections. The reconstruction test was used as before. The exposure time was fifteen seconds. The work con- ~ sisted in adding two columns of figures simultaneously. The total time elapsing between learning and reconstruction was three minutes. In one case the subject rested for three minutes. In the other case the subject worked immediately for one minute and then rested the last two minutes. This experiment should largely do away with the above objections. Four subjects did the experiments, the order of presentation being irregular in all cases. The averages are given in Table VI. TABLE VI Work Interval Rest Interval No. Records Subject Er. Time Er. Time Based Upon R.S. Wane 57.8 4.6 47.6 10 HG) ee ea ee de ol sere 12.0 37.0 3:45*, 34.0 9 Bee B Ox. car eee antares Byles capt Spilby byes 13 IDsS = Daas eee eee ee sate, See 3.6 47.2 10 As in the foregoing experiments, we find here a clear retro-. active effect for three subjects. E.B.S. alone shows no retro- action; in fact, the reverse is true. We can offer no explanation for this change excepting that through constant practice the subject learned the formations so well that no retroaction could operate. (B) There may be a further objection to our view that the work activity in some way actually is detrimental to the previous learn- ing. The introspections of our subjects indicated that many rest intervals were characterized by at least some consciousness of the original learning.. This consciousness would vary from a mere FURTHER STUDIES IN ‘RETROACTIVE INHIBITION 13 vague and fleeting thought of the learning to a vivid picture repeated a number of times. It may be claimed that the advan- tage of the rest interval lay in this very fact, namely, that during the rest the subject repeated the learning whereas during the work he could not. The point is a very crucial one. We have care- fully gone through the records of subjects F.C.D., I.D.S., and E.B.S., selecting, upon the basis of their introspections, those rest records which were free from any consciousness of the original TABLE VII CoMPARING AVERAGES OF SELECTED Rest INTERVALS WITH WorK INTERVALS AND witH TotTaL UNSELECTED Rest INTERVALS. RECONSTRUCTION TEST. THREE TRAINED SuBJECTS. MorNING AND EvEeNING ReEcorps. ONty Errors CONSIDERED. Morning Records . 0.5’ Interval 1.0’ Interval 2.0’ Interval Subject Rest Work Rest Work Rest Work PSDs abd cigs ney iosoo 8.70 4300) (11:20 8.70 12.1 N, 17 N, 8 N, 3 (4.2) (3.0) (6.3) Pears Sc tet wie) “OLGU 7.90 7.85 9.25 8257 hastoso N, 17 N, 14 N, 14 Oe35) (7.0) (8.70) BiG Dye Peek eee. 4520 8.30 2.00 8.30 ar 20 EeLOeo N, 5 N33 N,5 (4.3) (4.5) (5.4) Evening Hour ed Sie Me adler Bea ph LOU 7.60 4.83 7.80 3.86 9.70 N, 10 N, 6 N, 7 (1.95) (4.30) (3.20) Te eltoen Ma Bess «fos ots 20 ap 4.83 7.30 5.44 12.45 N, 10 N, 12 N, 9 (5.40) (4.75) (7.20) PO AE, Ve Ses 9 | 7.95 3.50 6.50 3.00 10.60 N, 4 N,8 N, 3 (3d) (2.9) (3.6) Expianatory Notes.—N represents the number of records upon which the average of the selected rest interval is based, The average given under Work is understood as based upon 20 records. The number in parentheses gives the average for the total unselected rest intervals as given in the preceding tables I and III. learning, with the exception of a short after-image which was always present. These three subjects’ records have been chosen because they furnish the most reliable and detailed introspections. Table VII gives the results in summary form. The data show that not only are these selected records confirmatory of our above stated view of retroaction, but the averages for the selected rest intervals are BETTER than the average for the whole series from 14 ERNEST BURTON SKAGGS which they are taken. We are not surprised at this, for we recall that the subject often said that the consciousness of the original material was vague, indefinite, and merely signified that he was not sure of the learning. While some of the return thoughts of. the original learning may have helped the subject in the rest. experiments, we believe that there is evidence that the rest period gains its advantage over and above this. | We believe that our data amply justify us in saying that atten- tive work, following the original learning of the reconstruction test, works in some positive way a clearly detrimental influence on the retention and recall of this original learning. Experiment Series B. Comparison of Retroactive Effects under Conditions of Fatigue Tolman (5) found evidence that retroaction is greater for evening than for morning conditions, suggesting that retroaction operates more clearly under conditions of fatigue. Following: are two studies which bear upon this problem: (1) Comparison of average error and retroaction in the initial and final experiments of a day's sitting. It will be recalled that at a given sitting six experiments were performed. We may assume that the subject was at least slightly fatigued at the end of the session. In the following tables (VIII and IX) we present the averages for the initial and final experi- ments for four trained subjects. Thus, e.g., our day’s experiments would be 1, 2, 3, 4, 5, 6—the odd numbers representing work, the even numbers rest, intervals. We thus compare 1 and 5 and TABLE VIII CoMPARING AVERAGE Errors AND RETROACTION IN First AND LAST EXPERIMENTS IN SERIES - Position 0.5’ Interval 1.0’ Interval 2.0’ Interval Subject inSeries W R W-R W R W-R W R W-R Evening Hour E.B:S. First 5.4 8.0 (-2.6) 8.0 5.0 (3.0) 12.0 86 (3.4) Last BS Zin edleihsiey 6.26.5) 0-13} 8.6 5.2 (3.4) 1 WIN Seno First 6.5 2.0 (4.5) i AN ee TM) Senn 24° Cie Last 10.41) 2.65)47:3) 12.1 41 (8.1) 13.6. 5.27) {648 F.C.D. First 8.0 6.0 (2.0) 8.0) 27a dd. Lb ae ast 10,5)" 8.5" (20) 13;2-"4,9" 4833) 17.5" 7.1" (i049) FURTHER STUDIES IN RETROACTIVE INHIBITION 15 2 and 6 in order to answer whether more errors are made in the final experiments of a day’s series. To find out whether retro- action is greater or less in the first or last part of the day’s series, we have taken the difference between 1 and 2 and compared it TABLE IX CoMPARING AVERAGE ERRORS AND RETROACTION IN First AND LAST EXPERIMENTS IN SERIES Position 0.5’ Interval 1.0’ Interval 2.0’ Interval Subject inSeries W R W-R W R W-R W R W-R Morning Hour E.B.S. First 74 46 (28) 11.9 7.0 (4.9) 11.9 11.2 (-.3) Last 8.68074. 8 Clk) 85 64 (2.1) 114 74 (4.0) EDS: First G16 14 a2) CLUE OLA MAL GO) TO:2p ois Coro) Last 13.0546:5)5.:(615) 133071267 (4.0) 13.4 64 (7.0) EG. D: First Ba 2 Coe) SRS eA GERD Bi7ou Ss ato) Last OS Ses CA) 10.1 6.0 (4.1) 17.1 6.7 (10.4) M.S. First 8.0 5.6 (2.4) 85:9 2:5)) 44 45 (-.1) Last 93 14 (7.9) 13.0 64 (6.6) ile tem G0)) with the difference between 5 and 6. The figures in parentheses give the retroaction; the other figures give the average error made. An examination of the preceding tables shows that the average error for both work and rest intervals is larger for the final experiments in a day’s sitting as compared with the average error for the initial experiments. Thus we find that out of twenty-one possible comparisons in our tables fifteen give a larger error for the final series, in case of the work intervals. Likewise, in case of the rest intervals, seventeen out of the possible twenty-one com- parisons indicate greater error for the final experiments. This may be due to fatigue or some intra-system inhibitory conditions which operate more as the experiments proceed. Both of these factors may work together. As regards the relative amount of retroaction, we find, compar- ing the figures in parentheses, that the greater retroaction occurs in the final experiments of the day’s series. Out of the twenty- one possible comparisons, fourteen give greater retroaction for the final experiments. In two comparisons there is apparently no difference. The facts seem fairly clear, indicating that in a learning process extending over about an hour’s duration there is more retroaction 16 ERNEST BURTON SKAGGS for the material learned near the end than for that learned near the beginning of the process. Doubtless we are dealing with a condition of accumulated fatigue plus a slackening of interest or incentives to do one’s best in the final experiments of the day’s sitting. Also there must be added a possible accumulative system of interferences due to the previous learning: Our subjects con- stantly reported that they could think of previous constructions and that it troubled them. However, until we disentangle the respective influences of the two above mentioned general factors, we do not feel safe in declaring that our findings support Tol- man’s conclusions. The presumption would seem to be that retroaction finds a fertile soil upon which to work in the case of fatigue condition or where the learning itself is poor. (2) Comparison of retroaction under morning and evening conditions. The second part of our present contribution to the question of the relation between retroaction and fatigue is based upon a comparison of retroaction under morning and evening conditions. Subjects F.C.D., I.D.S., and E.B.S. did evening experiments (7 to 9 p.m.) in addition to morning experiments (7 to 9 a.M.). The morning and evening experiments were carried out in alter- nating fashion, thus ruling out practice effects. No subject did morning and evening experiments on the same day. Our assump- tion is that the subject is more fatigued in the evening than in the morning—that morning represents a condition of relative freedom from fatigue. In order to decide whether there is greater retroaction present in the evening experiments certain conditions must be fulfilled: (1) The average error difference between the work and rest series in the morning must be LESS than the average error difference between the work and rest series in the evening; (2) the learning in the evening for the rest series and work series must be as good as that in the morning, otherwise the difference between morning and evening results may be due to difference in degree of learning in the morning and evening. Using a series of conventional symbols we may put the matter FURTHER STUDIES IN RETROACTIVE INHIBITION 17 thus: Let Er.W stand for the average error in the morning work series; let Er.R represent the average error in the morning rest series; let Er.w stand for the average error in the evening work series; and finally let Er.r signify the average error for the even- ing rest series. Then if there is greater retroaction in the evening, we have Er.W—Er.R < Er.w—Er.r, Granting, Er.r is not > Er.R The last condition is satisfied in our experiments. Much to the surprise of subjects E.B.S. and I.D.S., the evening learning was just as good as that of morning after the rest interval. F.C.D., however, demanded more exposure time to fulfil this condition and was accordingly given twenty seconds’ exposure instead of fifteen, as in the case of the other two subjects. TABLE X Subject: F.C.D. 4’ Interval: W-R= 8.3 -4.3 =4.0 Slightly greater R. in Evening w-r = 7.95-3.7 =4.25 (Diff.=0.25) 1’ Interval: W-R=10.1 4.55 =5.6 Greater R. in Morning w-r = 65 -2.9 =3.6 (Diff. = 2.0) Zaeiuterval iW —kR = 13.5) 946.1 Greater R. in Morning w-r 10.6 -3.6 =7.0 (Diff. = 1.1) Subject: E.B.S. 4! Interval: W-R= 7.9 -5.53=2.37 Greater R. in Morning w-r = 5.25-5.40—=-15 (Diff.=2.52—No R. in Evening) 1’ Interval: W-R= 9.25-7.0 =2.25 Greater R. in Evening wer = 7.3 -4.75=2.55 (Diff.=0.30) 2’ Interval: W-R= 15.5 -8.7 =6.8 Greater R. in Morning w-r = 12.45-7.2 =5.25 (Diff.=1.55) Subject: I.D.S. YZ’ Interval: W-R= 8.7 -4.2 =—4.5 Greater R. in Evening w-r = 7.6 -1.95=5.65 (Diff.=1.15) I’ Interval; W-R=11.2 -5.7 =5.5 Greater R. in Morning w-r = 78 -4.3 =3.5 (Diff. = 2.0) 2’ Interval: W-R=12.1 -6.3 =5.8 Greater R. in Evening w-r = 9.7 -3.2 =6.5 (Diff. =0.7) The data obtained indicate that for these three trained subjects there is little or nothing to make us believe that retroaction is greater in the evening. Subject F.C.D. shows slightly greater retroaction in the evening when using the half-minute interval, but for the one- and two-minute intervals there is greater retro- 18 ERNEST BURTON SKAGGS action in the morning. E.B.S. gets slightly more retroaction in the evening for the minute interval but greater retroaction in the morning for the other two intervals. I.D.S. alone gives evidence of greater retroaction in the evening, although in her case there is greater retroaction in the morning for the one-minute interval. Our subjects are too few in numbers to permit generalization, although for each individual we have tried to make the data con- clusive by securing a large number of records. As far as our evidence goes it is contrary to Tolman’s notion of greater retro- action in the evening. Moreover, our results make us suspicious that the facts indicated in section (1), preceding, are due to accumulative interference systems rather than to mere fatigue accumulation. The results of the section above and the preceding section seem in conflict. However, it must be remembered that in section (1) the subject was probably losing his incentives to learn, while in section (2), when working in the evening, the subject was thrown into a condition whereby he put forth extra effort to overcome his general fatigue condition. These differences in incentives and effort are, we believe, very complicating factors, and may explain the above mentioned differences. | Experiment Series C. Effect of Practice on Retroaction As one becomes more and more practiced with the learning material and the methods of learning that material, does retro- action tend to disappear? Robinson (4) found no retroaction in the case of one of his subjects who was very familiar with the chess board and chess men (reconstruction test): This suggested that the fact might be due to practice. We present the results for five trained subjects who have twenty experiments to their credit for each work and rest con- dition for each total interval used. We have taken the averages for the first half of the entire series and compared them with the averages for the last half of the series. The figures in paren- theses give the amount of retroaction. By comparing these figures for the first and last half of the total series we should gain FURTHER STUDIES IN. RETROACTIVE INHIBITION 19 some light upon the question of the relationship between practice and retroaction. The amount of retroaction should be less in the second half of the experiments in case practice decreases retro- action. It should be noted that subject F.C.D. has played chess and that E.B.S. and I.D.S. have played checkers a great deal. Consequently all knew the chess board well before beginning the experiments. TABLE XI COMPARING AVERAGES AND RETROACTION IN First AND SEconD HALF oF EXPERIMENTS Subject: F.C.D. 0.5’ 1.0’ 2.0’ 0.5/ 1.0’ 2.0’ Morning Evening Av. 1stl0(W) 8.9 13.9 15.9 13.0 133 Av. 1ist10(R) 4.4 (4.5) 3.9 (10) “i i (4) DG tas Pee Ae 2a 3.3) Av. 2nd10(W) 7.6 6.4 17.8 S45) 2.6 14.0 Av.2nd10(R) 4.2 (3.4) 5.1 (1.3) 5.7(12.1) 4.4 (1.1) 3.0 (-.4) 5.0 (9.0) TABLE XII COMPARING AVERAGES AND RETROACTION IN First AND SECOND HALF OF EXPERIMENTS Subject: M.S. 0.5/ 1.0’ 2.0’ Morning Av.1st10(W) 8.9 11.3 8.9 Av. Ist 10(R) ee aig Bape Av. 2nd 10(W) 9.7 SAS 9.9 Ay.2nd10(R) 3.5 (6.2) 8.9 (0.2) 9.5 (0.4) TABLE XIII CoMPARING AVERAGES AND RETROACTION IN First AND SECOND HALF OF EXPERIMENTS Subject: F.BI. 1.0’ 207 Morning Av. 1st 10(W) 8.3 7.3 Av. Ist 10(R) oF (7.0) a4 (5.6) Av. 2nd 10(W) Av. 2nd 10(R) 2 (68) 54 (05) TABLE XIV COMPARISON OF First AND SECOND HALF OF EXPERIMENTS Subject: I.D.S. 0.5’ 1.0/ 2.0/ 0.5’ 1.0’ 2.0% Morning Evening Av. 1st10(W) 10.3 9.9 127 his 5.0 Av. Ist 10(R) 2.9 (7.4) Fini eat8-2) 2.8 (4.7) ene 53.39) Av. 2nd 10(W) 7.1 8.0 Av. 2nd10(R)_ 5.4 (1.7) EB (66) 60 (5A) 1.1 (6.9) 42 (64) MM (9) 20 ERNEST BURTON SKAGGS TABLE XV COMPARISON OF First AND SECOND HALF or EXPERIMENTS Subject: E.B.S. 0.5/ 1.0’ 2.0’ 0.5’ 1.0’ 2.0’ Morning Evening 13.4 og. 6. 8.1 14.3 1.7) 8.6 (4.8) eres 6) 7.9 (-1) eG; 2) 9AtSia) oe 53 6.5 10.6 3.4) 5.4 (-3) 71 (43) Z9.(07) 1.2.45, 3) use oslo) Av.1st10(W) 85 Av.1st10(R) 68 Av. 2nd 10(W) 7.3 Av.2nd10(R) 3.9 ( ( If we take the total number of possible comparisons in Tables XI to XV, inclusive, we find twenty-three. Out of these twenty- three possible comparisons we find that eleven cases signify greater retroaction for the first half of the experimental series. Ten cases give more retroaction in the last half of the experiment series. Two cases are so close together that we may regard them as neutral. Looking at the comparisons by individuals, we note the following: Subject F.C.D. seems to reveal less retroaction as the experiments proceed, although in two comparisons the reverse is true. M.S. and F.BI. seem to indicate that retroaction decreases with practice. 1I.D.S. and E.B.S., however, indicate that retroaction increases as the subject goes on with the experi- ments. In the case of the former there are four comparisons giving greater retroaction in the last half as compared to two cases where the reverse is true. E.B.S. shows the same result. These results hardly justify us in making any definite statement as to the matter of the relationship between degree of retroaction and practice. The records of F.C.D., I.D.S., and E.B.S. are by far more significant than those of the other two subjects because they were thoroughly trained and have each six comparisons for inspection. Taking these three subjects, we find two experienc- ing greater retroaction in the last half of the work, while one does the opposite. Experiment Series D. Effect of Varying the Temporal Interpolation Muller and Pilzecker (3) and likewise Heine (2) found evi- dence that the sooner the interpolated work is introduced after the original learning the greater the inhibitory effects. Robin- FURTHER STUDIES IN RETROACTIVE INHIBITION 21 son,(4) however, comes to the opposite conclusion. The whole Perseveration Theory is largely at stake here, and the settlement of this point is vital to any theory of retroaction. In view of the conflicting results thus far reported it has seemed well to test further this point. Again the reconstruction test has been used as original learning material. The total length of the interval between learning and attempted reconstruction has been kept constant. The only vary- ing factor was the temporal position of the interpolated work. The experiments fall into two different series as follows (figures represent minutes in all cases) : Series I. Total interval 10 mintes: Method I. Learn—Work 3—Rest 7—Reconstruct. Method II. Learn—Rest 3—Work 3—Rest 4—Reconstruct. Method III. Learn—Rest 5—Work 3—Rest 2—Reconstruct. Method IV. Learn—Rest 7—Work 3—Reconstruct. Series II. Total interval 5 minutes: Method J. Learn—Work 2—Rest 3—Reconstruct. Method II. Learn—Rest 1—Work 2—Rest 2—Reconstruct. Method III. Learn—Rest 2—Work 2—Rest 1—Reconstruct. Method IV. Learn—Rest 3—Work 2—Reconstruct. Series II was done after series I because the first results were rather unsatisfactory so far as indicating anything definite was concerned. In the first series we used multiplication of two by four place numbers as the work. This work was unsatisfactory because it became automatic too readily and appealed very little to the interest of the subjects. Consequently in series II, in addition to reducing the total time to five minutes, algebra prob- lems took the place of the multiplication. These problems were simple equation problems with which the subject could make headway in the short work time. On any day’s sitting the subject was given all the various methods, there being a rest of from two to four minutes between each experiment. To equalize the fatigue factors the methods were rotated from day to day. In series I, E.B.S. alone did all four methods listed above; the others did three. ee ERNEST BURTON SKAGGS Nine trained and ten untrained subjects participated in this experiment. Tables XVI and XVII give the individual averages for the trained subjects for both series I and II. Table XVIII gives the averages for the untrained subjects. Graphs I and II will give the same facts for the trained subjects as found in their tables. SUMMARY TABLE XVI SHow1NG AVERAGES FOR FIVE TRAINED SUBJECTS. EFFECT OF VARYING REST INTERVAL BETWEEN L AND W, REcoNstRUCTION TEsT; SERIES I W3R7 R3W3R4 R7W3 R5W3R2 Subject Er, EB; Er. Ake ihn Ae Er ple Deus. 12.60 99.6 7.10 91.0 7.9 11.40 : Pe F.:G.D 8.47 58.5 8.00 51.9. 6.1 62.7 Be Ais I.D.S. 11.20 72.5 5.95 63.3 10.0 75.0 ier Ri ae M.S. 10.90 122.5 10.60 121.6 11.8 116.6 Bay, Weds FBS? 12.40 89.9 9.70 82.5 9.3 74.3 5.4 21.5 Total 55.57 443.0 41.35 410.2 45.1 442.6 5.4 ZL5 Mean 11.11 88.6 8.27 82.0 9.0 88.5 SUMMARY TABLE XVII SHOWING AVERAGES FoR Four TRAINED SuBJECTS. EFFECT OF VARYING REST INTERVAL BETWEEN L AND W, REcoNstRucTION TEsT, SERIES II W2R3 R1W2R2 R2W2R1 R3W2 Subject Er. ax Er. De Er. le Er. ‘ie F.BI. 3.13 2te 1.73 24.7 3.60 24.9 2.53 29.7 E.B.S. 12.00 734 7.30 61.6 8.87 66.6 6.00. 76.0 LDS: 9.87 62.0 8.00 56.7 7.47 52.8 8.87 58.7 R.S. 10.91 74.4 10.50 71.4 8.83 86.9 9.00 75.0 Total 35.91 230.7 27.53 214.4 287/23 kee 26.40 239.4 Mean 8.98 eer | 6.88 53.6 7.19 57.8 6.60 59.9 TABLE XVIII SHOWING AVERAGE TIME AND Errors FoR UNTRAINED SUBJECTS. EFFECT OF VARYING Rest INTERVAL BETWEEN L AND W. BASED UPON Four Recorps Eacu. Series II W2 R3 Rl W2 R2 R3 W2 Sex Er, Ae Er, as Er, AY Female..... 5.50 68.5 5.00 56.3 5.80 120.5 Female. . 1.50 44.0 4.75 36.3 ao 24.0 Females. ac 6.75 45.8 B25 42.5 2.00 47.8 Female..... 5.75 22.3 4.50 50.8 13.50 48.3 Female... 8.75 81.7 12.50 90.0 Pio 125.0 Female... 11.50 35.8 6.00 51.0 7S 42.5 Female..... 15.25 61.5 8.25 35.3 eye 507 Male visi tenucn 13.00 23.0 7.75 26.5 9.0 21.0 Males ch Ae t 2.00 49.0 2.00 52.0 2.0 65.0 Female... 3.20 41.6 1.60 25.8 4.6 64.0 Motal eer 73.2 473.2 57.60 466.5 71.15 608.8 Mean jy. cn uve s 7.32 47.3 5.76 46.7 732 60.9 MEY ETRE erty 3.84 14.3 2.29 13.5 3.83 26.2 FURTHER STUDIES IN RETROACTIVE INHIBITION 23 Turning first to the averages presented in the foregoing table for series I, we note that most errors are made when the work is introduced immediately after the original learning. However, comparing methods II and IV, we find that the former gives the less number of errors. We cannot account for this unless the work, coming at the end of the interval and thus just before the reconstruction, leaves the subject in an attitude which is unfavor- GRAPH I FROM TABLE XXXTV FRRORS WS5R7 RBW3 R4 R7 WS | Sih" ay aaipitdanmyangs Attssspmaaad pera bef oe 8S =aeEEw eee aew eges Ge Soe PE) Soi Vash aleiee raceme aie al elnie sie a Deemed a air ety so omoen 1p ores coe PED cl een eens gees ces GROUP AVE 24 ERNEST BURTON SKAGGS able to recall. Heine (2) found that this difference in attitude was insignificant in the case of nonsense syllables. Also, we have found very often that the subject was “ farther away ” after a rest than after a work interval, and our experiment testing the above assumption gave no evidence in its favor. Considering individual performances, which is more important GRY Str lt FROM TABLE XXxXW ERRORS W2 RS RIW2 R2 R2w2 Rl RSWS Faye ee GROVP AVE. FURTHER STUDIES IN RETROACTIVE INHIBITION 25 here perhaps, we find that F.C.D. gives a nicely graded series (see Graph 1). His results fit the perseveration view nicely. E.B.S. has a similar curve, although the differences are not so marked between the various methods. The other subjects all show more errors when the work is introduced immediately but, peculiarly enough, get their best results by the intermediate method. If we now inspect series I] (see Summary Table XVII and Graph II, above), we find, comparing the extreme methods, W2 R3 and R3 W2, that the group average shows more errors when the work is introduced immediately. However, we again fail to secure a graded series, reading from left to right in our summary table or graph, as we might expect if the perseveration view is correct and our methods adequate. A study of the indi- vidual records shows clearly that the temporal position of the interpolated work is significant—the “ work-at-once ”’ method is the poorest. A study of Table XVIII, giving the results for the untrained subjects, indicates likewise that the temporal position of the work is important. The most errors are made when the work is intro- duced immediately after the original learning. ’ Experiment Series E. The Degree of Similarity Between Inter- polated Work and Original Learning What is the relation between the degree of similarity of work interpolation and original learning as regards the amount of retroaction? Robinson (4) found evidence that retroaction is a function of the similarity between interpolated work and learning. Certainly the problem is a vital one from the standpoint of a theory of retroaction. If the work and original learning are identical, then we have mere repetition, and consequently mere reinforcement of the impression value and the associative bonds. But as the work becomes more and more dissimilar in content and method, where do we find a place where reinforcement stops and interference begins? Obviously two difficulties of experi- mentation appear. We must work out a priori a graded series 26 ERNEST BURTON SKAGGS of work activities ranging from identical to dissimilar. Also we must see to it that these various work activities are equally taxing, equally interesting and absorbing. Both desiderata are difficult to attain. In the following experiment series we have tried to obtain further facts upon this problem of degree of similarity. Two series of experiments were carried on (I and II below). As these two series differed considerably we will take them up separately : SERIES | The reconstruction test was used as in previous experiments. The variable factor was the content of the interpolated work. The following outline will give the plan of procedure: (a) S. studies the chess men arrangement for fifteen seconds. This we designate simply the original learning. (b) S. rests one minute. E. arranges the work series. (c) S. studies the interpolated work for fifteen seconds. The content of this work was the varying factor. (d) S. rests for one minute. (e) S. is tested for thirty seconds on the interpolated work or else continues the work if it is multiplication. In case of reconstruction type of work, S. tries to reconstruct. In case of picture study, S. tries to recall all details. If S. finished the reconstruction or the recall of the picture details before the thirty seconds were up (this seldom ever happened) he continued to think over what he had just done. (f) S, rests one minute. FE. records the score made on the work and makes ready for the reconstruction of the original learning. (g) S. tries to reconstruct the original chess formation. The work material consisted of four varieties for all subjects, excepting I.D.S. and E.B.S., as follows: I. Similar work. Here S. studied a new chess formation and after one minute tried to reconstruct it. FURTHER STUDIES IN RETROACTIVE INHIBITION 27 II. Intermediate form of work. A plain white cardboard was inked in heavy black lines such as to make sixty-four squares. This card exactly covered the chess board and was placed upon it during the work. Five articles were arranged upon these squares, namely, a large white button, large black button, a red checker, a black checker, and a pawn. _ S. stud- ied this formation and later tried to reconstruct it. Our assumption is that this work has some similarity to the original learning and yet differs in certain respects. III. Multiplication of two by four place numbers. Some sub- jects added double columns of figures instead, as the multi- plication became automatic. IV. Study of post card pictures of scenery. Here S. was told to study the pictures with the purpose of later recalling all possible details. These last two forms of work we have thought of as dissimilar and so signified in the following tables. Three to four minutes intervened between each experiment. A day’s experimentation included all four (or three, as the case might be) kinds of work. For I.D.S. and E.B.S. six chess men were used in the work series involving reconstruction work, whereas only five men were used for the other subjects. The order of presentation was rotated from day to day. The subject only rarely knew what kind of work was coming next. The tables giving averages for this experiment series follow (Tables XIX and XX). Table XLIII is a summary table for nine untrained subjects. If we will now compare the average error (Table XIX) for either of the dissimilar series (multiplication, or addition, on one hand, or picture study on the other hand) as over against the average error for the similar series, we note that the group means indicate that there is more retroaction where the work is similar. The reconstruction times show nothing. If we take the individual cases, as such, however, we find disagreement. If it is true that retroaction is a function of the degree of similarity, and if we really have three kinds of work differing in degree of 28 ERNEST BURTON SKAGGS SUMMARY TABLE XIX DEGREE OF SIMILARITY, SERIES I, RECONSTRUCTION TEST. TRAINED SUBJECTS III. Dissimilar IV. Dissimilar I. Similar II. Intermediate (Adding) (Picture) Subject Er. Time Er. Time Er. Time Er. Time BC.D; 10.17 49.7 3.42 47.8 5:93) 37.3 6.63 44.9 R.M.B. O70 62.4 11.90 66.4 9.80 91.9 15:38 65.4 LDS! 15.42 80.9 19s P5983 6.07 tetas ad AAP ee bo eae) Epis: ISSA S38 125: 03.2 9.83 74.1 She PLP ee R.S 15.80 65.6 11°70). 64a! 1OES0 a7 bo 13,5097 02.4 Total 65.26 312.1 46.19 300.8 48.73 322.1 $5.51) A724 Mean 13.05 62.4 P24" O02 9.75 64.5 741 34.5 ExpLanaTory Nore.—Averages given based upon 11 records for F.C.D., 10 for R.M.B., 12 for I.D.S., 12 for E.B.S., and 10 for R.S. The times of day when experiments were given eres 8:00 am., F.C.D.; 11:30 a.m. for R.M.B.; 7:00 p.m., I.D.S. and E.B.S.; 7:00 P.M., TABLE XX RECONSTRUCTION TEST. DEGREE OF SIMILARITY, SERIES I. SHOWING AVERAGES FOR 9 UNTRAINED SUBJECTS, BASED UPON 5 Recorps EACH Similar Intermediate Dissimilar Sex Er. Time Er. Time Er. Time Bemaleseeu. ds cisthl ataedaur Onl 53.6 12.2 81.0 4.0 59.4 Malem: cee een 80.2 9.8 42.8 Tome 36.2 Hemales gd eo ce ae 57.6 9.8 34.8 7.4 39.0 Female sinner el lce 61.6 6.2 56.4 9.8 56.2 Pemalesta. 3 eeine: peels 48.8 6.0 32.8 Ae 34.2 Bemalerse ao. ot vant a ae 0 93.5 £302 LOD ee 4.6 66.2 Femaleaay vier ote ie aoe 0) 50.5 9.0 40.5 Tht 32.6 Malele ec tene Mclean mene Ain Shoat TALE OL 6.5 2305 Female...... (if 55.4 4.6 39. A’.6 ie S5kG6 otal it Sat Pee ete ee LOR nS e7 78.35 464.9 62.1 382.9 Meat a Smads tress del tlle 4ter 9c 6 8.7 Si Ong 42.5 IME orc) RAE at iui Dies Ge Sine 2328 noid PRE OEORS similarity, then our summary table should show a decreasing average error as we read across from left to right. Method I, Similar, should give the greatest error; Method II, Intermediate, should give less errors; Method III, Dissimilar, should give least error. Unfortunately for such nicety of results, this is neither true for the group averages nor for individual means. F.C.D. shows least errors for the Intermediate method, although he makes most errors where the work is similar. I.D.S. and E.B.S. give results according to expectations stated above. R.M.B. gives a medley of results. He makes far more errors after studying the pictures than after any other kind of work; then in decreasing order of errors we find: Intermediate, Similar, and Dissimilar (adding). R.S. gives a similar irregularity of results. FURTHER STUDIES IN RETROACTIVE INHIBITION 29 Table XX, the summary table for the untrained subjects, indi- cates that the similar work causes greater interference than the dissimilar. In fact, their table gives a nicely graded series of errors, aS we might expect. This is true for both time and errors. Our data seem fairly indicative of the importance of the degree of similarity and in so far confirm Robinson’s results. SrERIEs II Feeling that there was the possibility that the different kinds of work used in the previous series might have involved different degrees of effort and attention, an additional series of experiments were done. Also we were urged to carry out these additional experiments to test further the matter of finding the point where similarity ceases to be conducive to retroaction and turns into actual facilitation. Consequently in these experiments we have confined ourselves entirely to the reconstruction test both for the original learning and the work. According to the decreasing degree of similarity as we deduced it, three different work arrangements were used, indicated in the following tables as I, II, and III. ‘Work formation I was very similar to the original arrangement excepting that the chess men were removed in any direction by one square (in a few cases by two squares). Work arrangement II involved the using of the same squares as in the original learning (thus keeping the same pattern) but different men on these squares. We have considered this arrangement as intermediate. Work formation III consisted in as widely dif- ferent arrangement of the chess men with regard to the original formation as was possible. Thus if the original formation had the men clustered together, in the work they were widely scattered. This we have considered as the least similar work arrangement. The procedure was similar to that in Series I. Five trained subjects did the experiments. On a given experiment day all three experiments were done twice, with an interval of from three to four minutes between each. Table XXI gives the individual averages. 30 ERNEST BURTON SKAGGS SUMMARY TABLE XXI DEGREE OF SIMILARITY, SERIES II, REcoNstRUCTION TEST. TRAINED SUBJECTS I vig III Subject Br: amen Er: Ww: Er, Time Er.W. Er. Time Er.W. BiG D2I GUE SF 6.7 140 118 10.6 518 14.0 R.M.B. 95 90.9 6.4 18.6 58.9 rps) 11.0 45.1 106 TDS: 5.8 74.6 oh 10.1 66.0 3.0 11:9)" 576 4.4 E.B:S: 8.3 42.0 Sal 5.9 41.9 6.7 a Lede eaeL 5.9 ieee 13.6 70.4 6.3 5.7 as.5 ed 124) <03.7 eee Total 40.4 273.0 26.6 47.0 239.6 34.1 58.5) 2915 east Mean 8.08 54.6 D2 94 479 682 117) 58.37 ee Norte.—Following are given the times of day and (in parentheses) the number of ips. vas veer sey Eel anne ¢ ee hig Coe (10) 5 GRAM. Bis 118A: M. (10); Taking the mean of the group of five subjects, we find that the just slightly dissimilar Series I gives the least mis- takes both for the reconstruction of the original learning (note column headed Er.) and for the reconstruction of the work series (see column headed Er.W.). Our so-called Intermediate Series II stands intermediate as regards both sets of error values. Series III, the most dissimilar series, gives MOST mistakes. It is of interest to note that there is an apparent relationship between mistakes in the original learning and the work series. A glance at the summary table shows that the smaller the error for the original learning, the smaller the error for the work, and vice versa. This may be indicative of interinhibitory or interference processes working backward and forward. Subjects F.C.D. and I.D.S. show a nicely increasing error value through Series I, II, and III. The increasing order of errors for R.M.B. is I, III, II; fOr HBS kle tills Oth ot LLALLe, Le From the standpoint of similarity of individual results our data are not satisfactory. The group mean, however, indicates that the most dissimilar work gives the greatest retroaction. Most of the subjects realized at once, in Work Series I, the fact that the change was slight. Here we approximated a mere repetition— at least definite relationship cues were present in the work arrange- ment which actually reinforced at times the original learning. Consequently we could hardly expect S. to make many mistakes. In Case II it was different. Although the same squares were covered as in the original learning, the chess men were rear- FURTHER STUDIES IN RETROACTIVE INHIBITION 31 ranged on these squares. This tended to bring confusion. The subject realized at once that he was dealing with the same pattern but the complete changing about of the chess men in some way brought confusion. In Case III the work arrangement was as different from the original as possible and so appreciated by the subject. The introspective aspect of this experiment series has been especially elucidating and suggestive. On the basis of these introspections and the above tables we may venture the following conclusions and interpretations: (1) Least errors were made in Work Series I. This was because the similarity approximated identity too closely, permit- ting helpful relationships to be easily noted. This is another way of saying that we had enough identical elements (largely in the form of spacial contiguity) to afford a positive transfer effect, however small. (2) The Method II was intermediate in errors. This was so because it stood intermediate between helpful similarity (as noted in (1) above) and harmful degree of dissimilarity, as was found in Method III. It offered a greater degree of dissimilarity than Method II, there being less chance to note helpful similarities and differences. On the other hand, there must have been some few helpful relationships in this method because it was not so destruc- tive of retention and recall of the original learning as was Method III. (3) Method III gave most errors. Here the point of helpful relationship between work and learning (at least the point of relatively harmless relationship) is definitely passed and the work acts detrimentally. (4) If we make up a series of apparently varying degrees of SIMILAR work, all series of which use the same material and methods of learning as the original learning, then the MORE DISSIMILAR the material the greater the retroaction. (5) We feel that the above is in no way contradictory to the facts found by Robinson or the results found in our preceding Series I. As Robinson has suggested, if we begin with original learning and work interpolation as identical, then there will be, 32 ERNEST BURTON SKAGGS of course, reinforcement, a mere repetition. As the work is made more and more different, there comes a time when the rein- forcing factors drop out or give way to interfering factors. Let us call this point, just mentioned, Point C. Up to Point C our statement above holds good. Here the greater the similarity the MORE the reinforcement and the LESS the interference. But as we go past this Point C the whole situation becomes reversed, and the more similar the material and method in the work the greater the retroaction, granting equality of attention and effort. This conception of the matter may be put in diagramatic form as shown in Diagram I. DIAGRAM I MAX} MUM AND INHIBITION AMOUNT oF REINFORCFMENT 2 = < k4 SCALE OF DEGREE OF SIMILARITY - Ficure 3. Explanatory Note—The above diagram is merely theoretical in its outline. Possibly the curves may be drawn with some mathematical precision in the future. The scale on the left vertical represents the amount of reinforcement and retroaction—two opposed processes. The horizontal scale represents the degree of similarity between the original learning and the interpolated work. Beginning at A where learning and work are identical, as. we go to the right there is greater and greater dissimilarity until at D the’ two are as dissimilar in content and method as is possible. At A inhibition is at a minimum and reinforcement at a maximum (mere repetition); at C the situation is reversed. At D the inhibition curve has fallen but never to the original minimum. Experiment Series F. Qualitative: Nature of the Rest and W ork Intervals In Part I, Experiment Section preceding, we noted that the validity of our results rested largely upon the nature of the work and rest intervals. However, it is extremely difficult to secure . | FURTHER STUDIES IN RETROACTIVE INHIBITION 33 many “ ideal” rest intervals—ideal in the sense that the subject was mentally passive and indifferent and entirely away from the original learning material. If one can secure ten such intervals out of twenty experiments, one is lucky. In a preceding section (see Table VII) we selected those experiments where the rest interval was characterized by practically no consciousness of the original learning after the after-image had faded away. No attention was paid as to whether or not the subject was “active ’’ mentally along other lines. A glance back at Table VII shows the proportion of “ ideal”’ rest intervals to the total of the rest intervals. In the case of the half-minute interval we note that there are 56 ideal rest intervals out of a total of 120; in the minute interval there are 51 ideal rest intervals out of 120; and in the two-minute interval there are 41 ideal rest intervals out of 120. Obviously, as the rest interval is lengthened there is a smaller proportion of “ideal” rest intervals. We found that ten- and fifteen-minute intervals were not only conducive to all sorts of mental activity on the part of our subjects, but they tended to put a premium upon thinking about the original learn- ing. For this reason we have not used long intervals in this investigation. | It must be remembered that an “ideal” rest interval meant no return to consciousness of the original learning. If we would include those intervals where the subject reported merely a vague and indefinite and fleeting consciousness of the learning, our number of “ideal ’’ rests would then be much larger. The cases where a complete “return”’ of the whole learning came to mind were very few. The consciousness of the original learning thus ranged from a mere fleeting and fragmentary visual or sub- vocal representation to a review of the whole situation. In general, we may say that as the subject progressed with the experiments he was better able to take an indifferent attitude during the rest intervals. There was always a memory after- image which faded out in the course of one to three seconds. All subjects reported at times that they felt that the learning material was “near at hand” and ready to “break in.” F.C.D. was the subvocalizer of the group. Often the subject would ’ 34 ERNEST BURTON SKAGGS report that a mere visual image of one or more chess men would flash into mind and out again without being connected with any chess board. At times a part of the chess board or the whole of it would flash into mind without any men placed upon it. Whether these fragmentary and isolated forms of consciousness were any aid toward retaining the original learning is difficult to say. We. would hazard a guess that in many cases they were not an aid but actually a hindrance. At times pictures of previous chess formations would come to mind, and most of the subjects felt (rightly or wrongly) that these were a hindrance to them. » In the work interval the subjects seldom ever reported any return to consciousness of the original learning. The after-image or memory after-image was usually cut short in the work inter- vals. If the work became automatic, then the subject tended to think of the learning or wander off into more interesting fields. On the whole our subjects reported that the work kept them busy. We may here anticipate the introspective results for the non- sense syllable experiments following by saying that the rest inter- vals were much freer from any consciousness of the original learning than was the case with the reconstruction test. The sense words were more inclined to return to consciousness than either the reconstruction test material or the nonsense syllables. ‘a> ae ae Pn OPE LM EN ATS OE OL LON SERIES OF TWELVE SENSE WORDS AS ORIGINAL LEARNING Will the same results found in the case of the reconstruction test likewise be found in the case of sense words? The following experimental section attempts to answer this question. In all the experiments reported in this section it is to be understood that the original learning consisted in a series of twelve sense words. The words were all one syllable, common words of three, four, and five letters (in a few cases six letters). The great majority of words were nouns, some verbs, and relatively few adjectives and adverbs. As far as possible the words in any one series were so chosen as to have little obvious relation between them. In all series the words were arranged in succes- sion, according to number of constituent letters, as follows: 4, 4, 4,5, 3, 4, 5, 5, 3, 4,5, 4. As can be seen, the four-letter words predominated.. The same instructions were given the subjects concerning the work and rest intervals as stated in the last section. Experiment Series A. Comparison of Work versus Rest Intervals For various reasons, as will be mentioned later, this series of experiments falls into three groups, (1), (2), (3). Group (1) Two trained subjects did these experiments. The sense words were printed in large black typewritten type on 3” x5” cards. They were exposed serially and at the rate of one per 1.5 second by moving a long covering cardboard, containing a small lateral window cut into it, down the card. A silent swinging pendulum beat the time. The words were exposed just once, as we desired to secure but a faint impression. The words having been exposed, there 36 ERNEST BURTON SKAGGS followed a work or a rest interval, as the case might be. The work consisted in adding ten one place numbers as fast and accurately as possible. One- and two-minute intervals were used. On any given experiment day six experiments were done, that is, three involving work and three involving rest intervals. Two minutes’ time elapsed between each experiment. Each day the order of the experiments was reversed, to-day being R, W, R, W, R, W, next time W, R, W, R, W, R. Each subject knew what kind of interval was coming next. The two subjects alternated in giving each other the tests, each subject doing his or her experiments every other day (9 A.M.). The following table gives the averages based upon twenty experiments for each work and rest interval, employing both one- and two-minute periods. Large R at the head of the column stands for the total number of words correctly recalled. Er. represents words recalled but which were incorrect. T signifies the time for recall. Figures in parentheses give mean variations. TABLE XXII One Minute Interval Two Minute Interval Work Rest Work Rest Roper Ls Re eee KRY? Bera: Re (hyo) LDS. 3.9004>.882 5.4 50.6) 828 3.0) 045.1970 7 Oe Caen) 24-0) eG pee) ON Ae ee OO ee BiB: 2.0 70.7 72.9. 2) Cine, AA eee (1. 1) (Oa) (18) (13) (08) (165) 43) 03) (12.9) (1.0) (0.5) (14.2) Inspection of the table shows clearly a retroaction in the case of LD.S. Subject E.B.S. indicates a slight tendency toward retroaction in the minute interval, whereas in.the two-minute interval no retroaction is found. Group (2) This group of experiments differs from the last in a number of respects. First, the order of presentation was irregular, and consequently S. did not know what interval was coming next. Second, only one interval was used, namely, a three-minute period. Third, S. worked only one minute, resting the last two ~ FURTHER STUDIES IN RETROACTIVE INHIBITION 37 minutes before recall. Fourth, two exposures were given instead of one. Fifth, the words were presented by a memory apparatus. The exposure apparatus used was a modification of the older Wirth machine made by Professor J. F. Shepard at the Univer- sity of Michigan. The series of sense words were written upon a strip of white paper in the same way as above mentioned. This strip was placed on the rotating drum of the machine and each word exposed serially at the exposure window. As there were fourteen stops during a complete drum rotation, and as we employed only twelve words, there was therefore an exposure of two blank spaces between the two exposures. The words were exposed serially at the rate of 1.5 second each, two exposures being given. The total exposure time was thus three seconds for each word. Each word was spoken aloud by S., who was instructed to take a discouraging attitude toward forming connections between the words. Other conditions were as in Group (1). Four trained subjects did these experiments. Table XXIII gives the means and mean variations (latter in parentheses). The averages are based upon nine experiments for each work and rest interval for all subjects excepting R.M.B., who has ten records. TRABLE XxX Work Rest Hour of Subject R Time Er. R Time Er. Experiment Shps. 7.0 77.4 0.3 yas Pane 0.78 11 a.m. (1.7) (30.4) (0.4) (1253) 01637 ee C0052) RS) 4.89 66.2 0.8 5.89 71.6 0.30 2:30 P.M. fey Cis.3)- (0,0) Cinco Los he COO.) R.M.B. Sed 82.5 0.2 4.89 ond 0.66 11 a.m. Creve (20) C0. 3) C196) eee 22015. C06) M.Bif. 5/8 48.6 0.8 6.10 50.3 0.78 1:30 p.m. Peso Cio, LO. 7) C1523) 016.2) (0:.69) Total PARSE Pa EN | ul ZAP 13 200"2 25, Mean 5.69 68.7 0.5 6.02 65.0 0.63 The above table indicates, in the case of three subjects, some retroaction. However, the difference between rest and work intervals was small, while in the case of R.M.B. the rest interval was more unfavorable to retention and recall. While we may say that some retroaction is indicated here, the results are not 38 ERNEST BURTON SKAGGS striking. The suggestion arose early that perhaps our work interval was too short. In consequence of the poor results obtained above, these experiments were discontinued and a third group begun. Group (3) The following experiments are exactly similar to the above in all respects except that the work was extended over two minutes instead of one. Thus we have two experiments : (a) The Work Series: Learn, work 2 min., rest 1 min., recall. (b) The Rest Series: Learn, rest entire 3 min., recall. The following table gives the means and mean variations for three trained subjects. All averages are based upon twelve experiments each for the work and rest conditions excepting in the case of F.C.D., who has fifteen records. TABLE XXIV Ly Work Rest Hour of Subject R Time ieler, R Time Er. Experiment F.C.D. Seek peu sD 0.33 6.07 56.5 0.2 8:20 A.M. - (1.51) (11.9). (0.53) (1.41) (12.8) (0.34) RS, 4.66 79.1 0.08 5.16 81.1 0.33 2:30 P.M. (1.56) (12.9) (0.15) (1.39) (8.81) (0.44) . M.BI. 4.92 63.1 0.5 6.17 63.8 1.0 1:30 P.M. (1.11) (10.3) (0.6) (1.19) (11.9) (0.8) In all cases in the table above we find that R for the rest condition is greater than R for the work condition. The rough time measurements signify nothing. Likewise a study of the errors made shows nothing definite. The experiments were also made upon five untrained subjects, students just finishing their first semester in the elementary labo- ratory course. They were given twelve experiments, six for each condition. Their averages are given in the following table. No mean variations are given. TABLE XXV Work Rest Hour of Subject R Time Er. R Time Er. Experiment R.M. 3.67 78.2 0.67 Seas 86.7 0.50 9 A.M. * H.G es, 84.7 0.17 7.00 67.0 0.30 4 P.M. idole: 5.58 92.8 0.25 5.58 80.2 0.40 11 A.M. G.J 5.00 69.9 0.67 7.00 69.0 0.67 1:30 P.M. E.D 3.20 43.4 0.00 2.80 49.0 0.20 4 P.M. Total. (22.78% 236920 1.76 25380 (B5109 2207 Mean 4.56 FS ie Mirela NS 5.14 70.4 0.42 i FURTHER STUDIES IN RETROACTIVE INHIBITION 39 Two subjects do better after the work, two better after the rest, while one does the same in each, as regards R. The aver- ages for the group indicate a retroactive inhibition. While these results are not convincing, they indicate, if we take the group average, some detrimental influence. Experiment Series B. Effect of Practice In the table below we have summarized the averages for the first and last halves of three subjects’ completed records. The results are taken from a study of the individual records whose means are given in Table XXIV, preceding. TABLE XXVI Position Work Rest Retroaction Subject in Series R R (Rest-Work) M.BIE£. First Half 4.83 474 0.37 Last Half 5.00 if Pe 2.20 F.C.D. First Half 4.86 526 0.74 Last Half 5.60 6.70 1.10 RS: First Half 5.50 4.83 —00.67 Last Half 3.83 5.50 1.67 If we measure the amount of retroaction by the formula, R rest minus R work gives the amount of retroaction (where R stands for the amount recalled), then we find, in the case of the above named subjects, that as they become more practiced in the tests retroaction increases. This is true in every case. As our practice period is at most rather short, we can only say that within the limits of practice involved in our present experiments the retroaction became more marked as the practice increased. This is possibly due to the fact that our subjects constantly improved in ability to take an indifferent and passive attitude in the rest period. Experiment Series C. Position of Experiment in a Day’s Experiment Series If we take the rest and work intervals which occur in the first part of the day’s experimental series and compare the retro- action thus found with the retroaction found between work and rest intervals occurring in the last part of the day’s series, what 40 ERNEST BURTON SKAGGS do we find? It will be remembered that a day’s experiment series consisted of six individual experiments. We have gone through the results for three subjects, whose means are given in Table XX VII, thus making the comparisons above stated. The following table gives the results: TABLE XXVII Position Retroaction Subject in Series R Work R Rest (Rrest-R work) BCs: First Half 5.78 O17 0.49 Last Half 4.67 5.94 1227, R.S. First Half 5.00 5.60 0.60 Last Half 4.20 4.86 0.66 M.BIf. First Half 4.70 7.00 2.30 Last Half Sea Ve anole 0.15 The data are not uniform in indication. In the case of F.C.D. there is more retroaction for those experiments occurring near the end of the day’s series. The difference is marked and cor- responds to his introspections where he believes that he gets worse as the experiments progress in a given sitting. The record for R.S. shows nothing. On the other hand, M.BIf.’s record shows that there is more retroaction for those experiments occurring at the beginning of the day’s series. About all we can say is that we have found individual variation in this regard. Experiment Series D. Effect of Temporal Interpolation The method of presentation and material were as stated in Experiment Series A, Group (1), with certain changes. A short exposure or learning was again used, the series of words being presented two times. The work consisted in solving simple equation algebra problems. The total length of time between learning and the attempted recall was kept constant in all cases, namely, five minutes. The variable condition is the temporal position of the work. Four different conditions were used, as follows: (1) \-R3-W2e.. Tage t. Sa ee eee a ee Recall (2), L-R2-W2-RI1 wih o, | iva ceeieas doe cae eeCeE (3.)“L-=RISW2-R2Z 0a i es ae, Recall LAYS Toa W cI Bud one ioithuss ins tad hs aie tine a ee ea mene Recall All four experiments were done at one sitting, four minutes elapsing between the end of one experiment and the beginning FURTHER STUDIES IN RETROACTIVE INHIBITION 41 of the next. The experiments were rotated from day to day. Excepting in the case of the last experiment of the day, the subject did not know, or very seldom, what kind of experiment was next. Unfortunately, only two subjects did this work, namely, I.D.S. and E.B.S., but for these two subjects the experiments were continued until their significance was clear. If the perseveration view is correct, and our work really involves a good type of concentrated activity, then we should theoretically expect to find least recall when the work followed immediately after the learning; then more recall for Method (3), more yet for Method (2), and most words recalled for Method (1), where we rest for three minutes after learning. The following table gives the results. The averages are based upon twelve experiments for each of the four different conditions. Experiments were done at 9 a.m. The figures in parentheses give the mean variations. TABLE XXVIII R3 W2 R2 W2 Ri R1 W2 R2 W2 R3 R Er. AX R Er. Ak R Er. Aly R Er. T EBS eae PO. Selo Sco Su)e Us comer ous D 3:33 0.08 272.6 PSO Oe Sore (0.65) (20.0) (1.38) (13.0); (1.67) (13.6) (1.0) (14.8) LDS: 3,06 el, 80903 4.08 1.00 86.1 4.08 1.30 88.3 4.8 0.83 74.7 (1.47) (21.8) (1.10) (17,4) C175) (28.3) (1.8) (18.7) Examination of these results shows diametrically opposed results. The record of E.B.S. is in accord with what we might theoretically expect from the standpoint of a perseveration view. The results of I.D.S. would indicate that she does best when the work is introduced immediately. However, the differences between the various methods are all very small and the safest thing to say is that this subject does not experience any greater retroaction when the work follows immediately. PART IV. EXPERIMENTAL SECTION NONSENSE SYLLABLES AS ORIGINAL LEARNING MATERIAL The following experiments attempt to give some light upon the same problems raised in the preceding experimental sections. However, in this section we have used nonsense syllables for the original learning. Nonsense syllables were used in one or the other of two forms, either as series of seven single syllables or as a series of seven paired syllables. Syllables having obvious meanings were thrown out. In any series we attempted to secure a balanced series of syllables with regard to difficulty of pronunci- ation. No two syllables in a given series began or ended with the same letter and no two syllables had two similar letters in order, as JEC, LEC. Experiment Series A. Using Single Syllable Series. Comparison of Rest versus Work Intervals The Shepard exposure apparatus, previously mentioned, was used. Seven nonsense syllables were typewritten in black large sized type upon a strip of white paper which was placed upon the drum of the exposure machine. As there were fourteen stops in a complete drum rotation, and inasmuch as we used only seven syllables, they were so arranged that every other stop of the drum gave a blank exposure. Each syllable was thus exposed for one second, then followed by a blank exposure; then the next syllable was exposed for one second, and so on. The tempo of the presentation was controlled by a silent swinging pendulum. The subject pronounced each syllable aloud once as it was presented. The syllables were exposed three times. After the exposure the experiment might take either of two forms, as follows: (1) a rest series of three minutes’ duration or (2) a work series, consisting of two minutes of concentrated work on mathematical or reasoning problems, such as the Thurs- FURTHER STUDIES IN RETROACTIVE INHIBITION 43 tone series A and B; then subject rested one minute. The minute of rest, following the work, was used to give the subject a chance to rest and also to get away from the “ work attitude.” The subject then tried to recall all syllables possible. At one sitting six experiments were performed, three involving work and three involving rest intervals. Between two and three minutes elapsed between each experiment. Subjects I1.D.S. and E.B.S. alternated every other night as subjects in these experi- ments. The other subjects performed twice a week. The following tables give the averages for five trained subjects and fourteen untrained subjects. R stands for the number of correct reproductions; Er. for the errors; and T for the time of reproduction (taken with a stop watch). Half credit was given if two letters, in their proper sequence, were reproduced, as HIG for LIG. The time represents the interval between the signal for recall and the final statement of the subject that no more syllables would come (S. was asked not to “rack his brain”’ for the syllables). Again introspections have been carefully taken. TABLE XXIX SHOWING AVERAGES AND MEAN VARIATIONS FOR FIVE TRAINED SUBJECTS. Series A, SINGLE SERIES Non-SENSE SELLABLES. REST vs. Work INTERVALS Work Interval Rest Interval Subject R ai: Er. R Tr Er. Biss aol 24 0.54 4.23 82.2 0.54 (1.15) (16.1) (0.48) (0:66) © C1837)" 0s} E.B.S. Zrl3 64.1 0.27 2:37 90.9 0.43 (0.88) (16.2) (0.39) (0.96) (22.8) (0.52) F.C: 3.67 63.1 0.60 4.70 76.8 0.43 AOrrao wt G18}9)> 9/(0..52) (i260 )s (2234 ees) M.BIf. 2.92 45.8 0.83 3.88 Sb 5 0.41 Clyi7) 9a y7 (0.72) (0.75) (17.8) (0.49) Sw. 3.75 61.0 0.67 4.30 68.0 0.42 (0°79 )o 0 Cla. aya OC 0,61) CE .00) = 613, OR const Notze.—R stands for number of syllables recalled; T, for time in seconds; Er., for errors. Figures in parentheses stand for mean variation. The times of day when each subject did the experiments and the number of records upon which the average is based are as follows: -D.S., 7 P.M., 14 records; E.B.S., 7 v.m., 15. records; F.C.D., 8 am,, 15 records; M.BIlf., 1:30 p.m., 12 records; and Sw., 2:30 p.m., 12 records. 44 ERNEST BURTON SKAGGS TABLE XXX Work Rest Sex R. Er ay R. ley “Ls Female iit eas 4.08 0.08 66.7 4.60 0.42 Gh 4 Female’ scinsaiacatisietsste 2.92 1.08 122.0 Shay 1,83.) 13034 Femaleticinica nanan 3.67 10041001326 BID 0.33 91.8 Females esc eiai ees 3242 0.25 68.7 3.92 0.42 80.2 Hemalece teen piaiine 3.42 0.25 56.3 3.43 1225 SleSe Bemalets ae cache ations 1275 202m 10652 3.92 19925 22208:.0 M ale a oo a URN peice nan 2.75 0825 59.0 Zo 0.50 69.8 Male Wht Aunt tee ve Gren 5.33 0.00 54.2 nes 0.30 86.0 Malem rae raiies setae g 2107 0.00 46.0 HM oe NUR, 41.2 Male unc maneneeen els 4.82 0.70 67.0 5e50) 0.30 69.0 Maile se, nue ned oo Z10 0.75 Ties 3207: 0:60! ant 7/988 Male ett ena s snare ea 2.83 0.33 57.2 3.25 0.25 520 Male Raat ieee 3.75 1.10 0747, 4.50 0.50: sl07e3 Maleate he tray ug 4.00 0.50 TZ 5.60 0.10 62.3 Lotaliis seine mena eo 7.21 1048.8 57.48 8.89 1083.6 IY Dee are heer a hele i Oe 3.39 0.52 74.9 4.10 0.64 77.4 MEV Re io Arne OEE ts 0.71 0.79 Table gives averages for 14 untrained subjects (elementary psychology students). Non-sense syllables, Series A. Experiments done at various times of day from 10 aM. nntil 5 p.m. Each average based upon six experiments, total experiments being 12. A study of Table XXIX (trained subjects) shows in every case that more syllables are recalled after the REST interval. While the difference between the two methods is indeed slight, yet it is consistent and argues for retroaction. The error columns show nothing significant. As regards time of recall, every sub- ject takes longer after the rest interval. The only plausible explanation we can offer for this is that after the work the syl- lables are gone so completely that, having reproduced a given number, the subject realizes the hopelessness of further attempt. Again, differences in attitude on the part of the subjects may possibly explain the situation. Table XXX gives the averages for fourteen untrained subjects, based upon six experiments for each of the two kinds of intervals. The students were in the writer’s elementary laboratory section, just beginning their laboratory work in psychology, and wholly ignorant of the purpose of the experiment. The averages for these untrained subjects indicate quite clearly retroactive inhibition. Taking the individual cases, we find that twelve out of fourteen show such retroaction. In one case the method used does not seem to matter, while in one other case the subject does better after the work interval. FURTHER STUDIES IN RETROACTIVE INHIBITION 45 On the whole, the above experiments indicate clearly, we believe, a retroactive inhibition, a detrimental influence exerted by vigorous attentive work upon the original learning. Experiment Series B. Using Method of Paired Associates Testing (1) Rest versus Work Intervals. (2) Similarity of Work and Learning. In the following reported experimental section we have tried to get further data on (1) the comparison of rest and work inter- vals, and (2) by using relatively similar and dissimilar material in the work we have tried to find which type of work brings about the greater inhibition. In these experiments we have used two criteria for inhibition, namely, reaction times and the amount reproduced. Apparatus and procedure: The exposure apparatus previously mentioned was used, with a reaction time addition. An Hipp, spring driven, chronoscope was used for measuring the reaction times. It was used on a simple make-break circuit, the “ make ”’ occurring as the test syllable was shown in the exposure window and the “ break’ occurring as the subject released the teeth key. The chronoscope was kept tested by means of a gravity chrono- scope in which a steel ball, falling through one meter distance, operated the Hipp on a make-break arrangement. Our chrono- scope has been fairly constant for our purposes, having a constant error of 28.3 sigma, + 4.6 sigma. Seven pairs of syllables were exposed to the subject fifteen times by means of the rotating drum and control lever. A pause of two seconds occurred after the eighth exposure. Each pair of syllables was exposed for one and a half seconds. A blank space was shown at the window between each syllable exposure. The subject pronounced with equal emphasis each pair of syllables as it appeared. Having given the fifteen exposures, any one of the following three procedures might be followed: I. A rest period of five minutes. Test. II. A work interval using relatively dissimilar learning mate- rial. Here the work consisted either in working simple 46 ERNEST BURTON SKAGGS arithmetic or algebra problems or else in reasoning out the Thurstone reasoning problems. The work lasted for three minutes, then a rest of two minutes ensued. Test. III. A work interval using relatively similar material. S. studied another set of seven syllables for one minute; worked for one and one-half minutes on the same material as given in II above; recalled syllables from the work series for one-half minute; rested two minutes. Test. As will be seen, the total time between learning and testing is in all cases five minutes. In the test for recall the syllables con- stituting the left side of the pair (from the subject’s side) were presented and in the following sequence: 5, 7, 1, 3, 6, 2, 4. (These numbers refer to the position of the syllables in the original learning.) The subject reacted by speaking aloud the associate syllable as soon as the syllable came to mind, or in case nothing came, the subject signified by saying “nothing.” In either case the speaking broke the electrical contact and +e the chronoscope. The following tables give the results in terms of reaction times and number of associates reproduced, along with errors. We have treated the reaction times in three ways: (1) We have picked out from our records the wholly correct associates and thus compared the three methods; (2) we have combined the wholly right and half right associates; (3) all association times have been lumped together for each method used and thus com- pared. iis thus include right, half right, errors, and reaction times giving “ nothing.” If there is a positive retroaction it ought to be manifest in a comparison of Method I with either of the other two methods, II and III. We should expect the shortest reaction times for Method I, involving the rest interval. These results are found for only two subjects, F.C.D. and I.D.S., in the case of the right associates. For the other three subjects the data are conflicting— at least not clear. If we consider the times for correct and half correct associates, we find no evidence for retroaction in the case of any subject. Taking the association times for all reactions a; FURTHER STUDIES IN RETROACTIVE INHIBITION 47 TABLE XXXI REACTION T1MEs, FivE TRAINED SUBJECTS Times for Correct Times for Correct Times for All Associates and Half Correct Reactions Subject I II III I II Wy I II Ill F.C.D. 3594 4664 4300 4782 4935 4320 6931 7398 5949 (3126) (3445) (2839) (4104) (3423) (2736) (5929) (5550) (3919) N,31 N,35 N,37, N39 N34 N56 N62 N,54 LD.S. 2443 2358 3840 2619 2419 3283 3686 3829 4605 (1362) (1244) (1757) (1555) (1265) (1766) (2312) (2155) (2078) N,43. N41. N39 N48 N,43. N42 N,79 N82 N82 E.B.S. 3333. 2790 «-.2734:«=S3688.-«2799:« 2907» 4801 4696 4489 (1856) (1272) (1271) (1962) (1195) (....) (2280) (1918) (1945) N,29 N,25 N32. N33 N26 N38 N,83 N80 N,79 M.BIf. 3074 2866 3490 3314 2785 3460 4378 4062 5029 (1225) (1220) (2100) (1474) (1163) (1970) (2342) (2615) (2510) ‘N,23. N28 N,19 N,31 N30 N,25 N56 N54 N,55 Swk. 3888 4689 3669 3972 4526 3800 6530 7756 7188 (2817) (3117) (1791) (2697) (2993) (2283) (4778) (5686) (4847) 2 N,25 N,30 N,24 N30 N33. N,26 N63 N,55 N,52 roup Deans) 93260)" 3474013547.) 3675.» 3493) (28572) 5283) 554865452 In the table above the four place figures give the reaction times in sigma and are averages. The figures-in parentheses are mean variations. N stands for the number of cases. Column headed I refers to Rest Interval; II, the Work Interval using Dissimilar work material; III, the Work Interval involving Similar work material. TABLE XXXII Amount, RECALLED AND Errors, FivE TRAINED SUBJECTS I. Rest II. Work, Dissimilar III. Work, Similar Subject R Er. R Er. R Er. E.B.S. 2.83 0.70 3.08 0.46 2525 0.67 (1.08) (0.74) (1.24) (0.38) (0.75) (0.38) TDS: 4.20 1.30 BisfAl 0.79 e/5 Tal (0.99) (0.73) (1.47) (0.45) (1.00) (0.64) Swk. aal5 1.65 4.55 1.39 3.83 1.00 (0.80) (0.85) (1.01) (0.77) (1.41) (1.00) M.BIf. 3.50 1.89 3.88 E25 2.88 2/5 (0.63) (0.57) (0.78) (0.94) (0.47) (0.75) GD: 4.00 1.60 4.22 1.00 4.25 1R62 (1.00) (0.75) (0.96) (0.66) (0.94) (1.03) Group Mean 3.54 1.43 3.89 0.98 3.39 1.44 Figures given in table are averages. Figures in parentheses give the mean variations. under each method, we again find our results conflicting. Subjects Swk. and I.D.S. alone give data suggesting retroaction. Table XXXII gives the amount of recall and errors under the headings R and Er., respectively. Comparing the results again as regards the rest as over against the two work intervals, we find that only one subject, I.D.S., indicates retroaction. Likewise the 48 ERNEST BURTON SKAGGS group average signifies nothing. A study of the errors made also seems fruitless. Our work (repeating in many ways the work of DeCamp) with the paired associates method has been largely barren of results. Strangely enough, we secure evidence of retroaction for single series of syllables, but our results are conflicting and often the very reverse of expectation when we used the paired associates material. Our only explanation is that fifteen exposures in this case fixes the associations too securely to permit retroaction. However, this hypothesis hardly seems justifiable as we review the struggles of the subject to recall the associate. A series of experiments must be carried on in which the number of exposures is greatly reduced in order to test this point. The second problem investigated in this section has to do with the question of the degree of similarity of the work and learning material. A review of the two preceding tables gives the results for five trained subjects. The reaction times for Methods I and II (Table XX XI) show only two cases, when right associates are taken, where the similar work material exerts a greater retroaction than the dissimilar. For the other four subjects the reverse is true. The other studies of the times in the table are likewise conflicting. If the more similar work material exerts greater retroaction than the dissimilar, then we might expect a longer reaction time in the former case. Our results do not indicate that such is the case under the conditions of our experiment. If we turn to the errors and amount recalled (Table XXXII) we find some evidence that the more similar work exerts greater retroaction than the less similar. In three cases R is less for the similar work than for the dissimilar, and in the two cases where the reverse is found the difference is very small. Taking the group averages, there is a slight advantage in the case of the dissimilar work. If we overlook the fact that our reaction times failed to reveal anything, our evidence indicates that the more similar the work and learning, the greater is the retroactive effect. FURTHER STUDIES IN RETROACTIVE INHIBITION 49 Experiment Series C. Single Syllables Similar versus Dissimilar Work In the following section we have used eight untrained subjects and single series of nonsense syllables. The series of seven syllables was exposed on the rotating drum at the rate of one per 1.2 seconds. Three exposures were given, the subject saying aloud each syllable as it came to view. After the exposure S. either worked for three minutes on Thurstone reasoning prob- lems or else worked for two and one-half minutes studying fourteen new syllables, taking the last half minute to recall these syllables. Six experiments were done at one sitting. Each sub- ject did his experiments at the same time of day, although different subjects did their work at different times. There were four men and four women, all elementary psy- chology students having had some laboratory work but no previ- ous work with nonsense syllables. They merely knew the general nature of the experiment, namely, that it had to do with inhibition. In the recall the first syllable was always given to the subject. A perfect recall would thus be six syllables. The time taken in recall was measured by a stop watch. The following table, XXXIII, gives the individual and group averages : TABLE XXXIII Similar Dissimilar R Br T R Ds 5 ah 0.83 1.70 104.8 1.88 1.50 15120 175 0.25 45.2 2.92 0.25 48.3 0.90 0.50 88.2 2.60 0.50 116.4 1.92 0.75 92.5 2.00 0.67 92.3 1.00 0.83 46.5 tego (aos 75.0 126 7eeee Vl 500 insteds Ze25 1.08 116.7 Beas 0.50 128.7 3.83 0.50 141.3 2.08 5.83 103.0 Dros st gly 140.7 Total 12.48 11.36 622.4 19.14 6.00 881.7 Mean 1.56 1.42 77.8 2.36 EVE: LOR? M.V. 0.48 0.54 Without exception, every subject recalled more syllables when the work was dissimilar. Likewise the errors made are greater in the case of the similar work. The similar work apparently involves a greater retroaction. 50 ERNEST BURTON SKAGGS Experiment Series D. Effect of Varying Temporal Position of Interpolated Work, Single Syllables In the following report series of seven nonsense syllables were used as the original learning material. The method of exposing and learning was similar to that in Experiment Series C. The interpolated work consisted in working algebra and arithmetic problems, and, as such, was a constant condition. The con- dition which we have tried to vary is the length of time elapsing between original learning and work interpolation. The total time was the same for each experiment, namely, six minutes. Three variant intervals or combinations were used, as follows (figures indicating minutes) : I, L—W3—R3—Test. Il. L—R1—W3—R2—Test. Ill, L—R3—W3—Test. On any experiment day six experiments were done; that is, each of the above three was done twice. An interval of from two to three minutes intervened between each experiment. In the test E. gave S. in every case the first syllable. Thus there remained six syllables to be reproduced. Half credit was given for those syllables in which two consecutive letters were correct. In the following table, XXXIV, are given the results from five trained subjects. The column headed R gives the average amount reproduced (six being a perfect score); Er. stands for the average error; and T signifies the time taken in recall. Unfortunately, the experiments had to be done at different times of day. E.B.S. and I.D.S. did their experiments at 7 pP.M.; F.C.D., at 8 a.m.; Swk., at 2:30 p.m.; and BIf., at 10 a.m. Taking first the group average, it will be noted that, while the difference is not great, there is a better recall when a rest interval of three minutes intervenes than when the work is introduced immediately. The intermediate combination II, however, gives’ the best results. The averages for this intermediate method are based upon relatively few records (unfortunately) in the case of three subjects. Studying the data by subjects, we note that I.D.S: and E.B.S. FURTHER STUDIES IN RETROACTIVE INHIBITION 51 give rather clear results, a decreasing value of R as we read across the table from left to right. F.C.D. and Swk., while doing better (apparently) by the Method II, yet indicate in their averages that there is more retroaction when the work is intro- duced immediately than when a rest of three minutes intervenes. Blf. alone fails to indicate a greater retroaction when the work is introduced immediately. On the whole the data of Table XXXIV indicate that work introduced immediately after learning gives the greatest retroaction. As a check series some further experiments were done with thirteen untrained subjects. These experiments were similar to the above excepting that the total time was seven minutes rather TABLE XXXIV L-R3-W3 L-R1-W3-R2 L-W3-R3 Subject R Er. a8 R Er. Ai R lie fh E.B.S. 2.04 0.80 97.3 145 0.45 82.5 1.30 0.62 70.4 (0.97) (0.63) (16.8) (1.04) (0.58) (20.5) (0.10) (0.50) (25.4) Dae ey rae 2 NTE ONE oe tg N IZ GN IZ NZ LDS. 3.497 0.33." 63.0 3.10 0.40 65.4 2.63 0.88 64.0 (1.04) (0.44) (20.0) (1.22) (0.40) (17.7) (1.15) (0.86) (20.7) Pyec te, NZ ON IO NAG CN LOPS N 12 NZ Ne FED; axa08: 1.600 -72.0 5.20 0.20 59.0 295 0.55 71.0 (1.24) (0.46) (25.0) (0.32) (0.32) (22.0) (1.17) (0.54) (19.0) N,10 N10 N,9 Nios SSRN N,10 N,10 N,10 Swk. Soe. Won 77.1 4.67 0.00 64.5 2.85 1.05 100.3 (0.67) (0.36) (22.9) (0.89) (0.00) (19.5) (1.05) (0.47) (22.9) N,10 N,10 N,10 N6 N6 N,6 N,10 N,10 N,10 Bif. Boe as. | 94.6 430 086 41.7 4.20 040 51.5 (1.04) (0.68) (10.6) (0.33) (0.49) (6.6) (1.04) (0.32) (6.8) N,10 N,10 N,10 Res Beg ds iar N,10 N,10 N,10 Group Aver. 3.22 0.58 73.8 3.75 0.38 62.6 2.792 0 707 A Figures in parentheses give M.V. WN signifies number of experiments. than six. In all cases S. spent the last minute just before recall in glancing through a magazine. Only two combinations were used, the I and III above. Table XXXV gives the results. The averages are based upon only four experiments for each combination. The work con- sisted, on the first experiment day, in working out Thurstone reasoning problems and, on the second day, doing simple algebra and arithmetic problems. The above table indicates nothing definite. The group aver- ages indicate a slightly greater recall when a rest period inter- De ERNEST BURTON SKAGGS TABLE XXXV R2 W3 M1 W3 R2 M1 Sex R Er. ah R Er. Az Male iets Ue ears S355) 0.75 O1eZ Shs 0.75 58.2 Males cee ee ia e218 0.50 70.2 Pas 1.00 68.5 Malerget cere 1.6 0.88 Fis 0.8 2.00 81.0 Male.... ath 235 1.50 78.2 $395, 1.12 108.0 Female. . 2.9 1.38 98.2 3.6 12412 JaaG Male.... BS 0.50 40.9 2.5 0.50 40375 Male trivetnicsc ays 0.62 41.5 Sioa! OF38 822 Female.... 40) 0.88 104.7 3.0 0.38 9725 Malet aineias 2.6 0.12 Sie/ he: 0225 Sono Malepeua. 2ro 0.25 48 .2 BASS 0.25 49.0 Female bir)... Shel 1.10 74.8 2.0 0:50). 1012 Pemalesiay 2) 4.2 0.88 45.0 Say 1.00 56.7 Male taiynay. 223 0.75 47.0 3 1225 S162 Total, eee kien cid ot OLA Pe Led e757 2 35.4) 1 0u50 Msooes Mitel Rik, aitedleaech cee wea, 28 OD tet OOO 67.3 PAVE NU EAW bY) iS. Mr Nee Pepa dt crete eects aa GO PMOL 18.6 0273000241 20.2 venes before the work, as compared to the condition where the work comes immediately. Seven individuals show greater retro- action when the work comes immediately, while six give opposite results. Basing our statement on the work of the trained subjects, we may say that there is a positive indication that the temporal position of the work is important. Experiment Series E. Effect of Varying Temporal Position of Work, Paired Associates Method The problem investigated in this series is the same as that considered in the Series D preceding. However, we have here tried to test the matter of the importance of the temporal position by using the paired-associates method. The arrangement of the pairs, their presentation, and method of testing was the same as mentioned in Series B preceding. Two different methods were used and compared, as follows: I. L (15 exposures )—W3—R3—Catalogue 1—Test. II. L (15 exposures )—R3—-W3—Catalogue 1—Test. In each of the above methods the subject spent the last minute glancing at the pages of an apparatus catalogue. The records of five trained subjects appear in the following tables, XXXVI and FURTHER STUDIES IN RETROACTIVE INHIBITION 53 XXXVII. Three experiments were done at each sitting, two of one method and one of the other. On the day following, this was reversed. The subjects did their experiments at the same times as mentioned in Series B. The presentation of the methods was irregular and S. did not know what to expect next. Table XXXVI gives the means, the mean variations (figures in paren- theses), and the number of experiments upon which the means are based. These means represent the average number of syl- lables (associates) reproduced correctly (column headed R), and the errors (column headed Er.). Table XXXVII gives the reaction times in sigma, the mean variations, and the number of cases. TABLE XXXVI : R3 W3 Cl W3 R3 Cl Subject R Er. R Er. LE 0 ED rly A? SG Ea 4.07 1d 4.36 1.07 R123) (0.80) (0.77) (0.80) N,7 N,7 N,7 N,7 MESO cas ine es cca ee ree. 3.61 1.94 255 2.67 (1 ve ee (1.42) (0.99) $ ’ 9 > Ge RS A ee ee 5.00 1.00 4.11 1.00 (1.22) (0.55) (1.12) (0.67) 9 N,9 9 : EAL SSE RE 1 ls a abe 28 2.81 0.89 2251 0.96 (0.98) (0.46) (1.30) (0.88) N,13 N,13 N,13 N,13 TULD anes fF Circa y 6a Oso oie eae 4.40 1.50 4.80 1.00 (1.30) (0.83) (0.73) (0.33) N,l1 N,11 N,12 N,12 GroupeVlcanedy.). «2. cikvelne o/s 3.98 1.38 3.67 1237 Studying Table XXXVI, we note that the group average indi- cates an advantage with regard to amount reproduced for the method in which a rest interval of three minutes intervenes between the original learning and work. The errors are the same. Three subjects give more associates when the work is delayed, whereas in two cases (I.D.S. and Swk.) there is a slight advantage for the method which involves the immediate introduction of the work. Turning now to the reaction times (Table XXXVII) and inspecting first the group average, we note, for the correct asso- ciates, that the mean reaction time is 125 sigma longer when the 54 ERNEST BURTON SKAGGS work is introduced immediately. Considering columns three and four in the table, which give the times for both correct and half correct associates reproduced, we find the same thing, excepting that here the difference is greater, namely, 369 sigma. It will be noted that in the case of Swk. we have used a smaller set of means than that listed above his mean variations. We have taken out ~ three very extreme times from his records and this gave us the smaller averages. These extreme records were thrown out as TABLE XXXVII Reaction Times Reaction Times for Wholly for Correct and Half Correct Associates Correct Associates Subject R3 W3 Cl W3R3 Cl R3 W3 C1 W3R3 C1 Sik Peete aahte tater eae 6026 3965 5564 5375 (4921) (2557) (4268) (3516) N,24 N,27 N,30 N,34 *3967 3472 3702 3418 Hey hit Rep el Reg WERE a Rhee Le iy 2701 2465 2699 2903 (1401) (1248) (1327) (1739) N,30 N,20 N,33 N,26 | WN CEI Ds Batre syed Pie ai HAS A He 4149 4580 4111 5236 (2898 ) (3081) (2768) (3709) N,43 N,3 N4 537, FYB AS 2 si enbhorie durante 2775 2368 2941 MPAIER. (1496) (1408) (1583) (1381) N,34 N,32 N,37 N,33 6B eS aa) A 1 AL Si 2327 3660 2447 3854 (1253) (2334) (1370) (2502) N,43 N,54 N,48 N,62 Gr OUD iM Caney hace isos «suit 3184 3309 3180 3549 * This row of means used in all calculations. the subject reported that he had the associate but held on to it overtime. Consequently we have used his smaller average, in which these three very extreme reaction times are omitted from consideration. Subject Swk. gives a reaction time longer for Method II, although this difference stands out greater than the difference between the two methods as regards amount reproduced. About the same can be said for E.B.S., his reaction times being shorter . when he works immediately after the learning. F.BIf. gives a shorter reaction time for Method I (immediate work) in case of correct associates. However, when we consider together the correct and half correct associates given the reverse is. true. We FURTHER STUDIES IN RETROACTIVE INHIBITION 55 are inclined to stress columns III and IV in our table above for it is these half correct associates which represent the hesitations and blockings noted in the subject’s reactions. Considered from the standpoint of either correct or correct and half correct asso- ciates, the results from F.C.D. show a longer reaction time when the work follows immediately. While, as we saw above, I.D.S. reproduces as many associates for Method I as for Method II, a study of the reaction times gives a clear case for Method II. The reaction times are longer when the work is introduced immediately. We believe that a careful analysis of the above results justifies us in saying that the temporal position is important—that work introduced immediately causes more retroaction (as judged by ‘a longer reaction time and a smaller number of recalled associates ) than the method wherein the work is not introduced until after three minutes. We have some individual (apparent) exceptions. Some other form of work activity, a longer period of work activity, less number of exposures in the original learning—these and other differing conditions might have yielded us, even for these apparent exceptions, results confirmatory of our above generalization. Certainly we must say that there seem to be individual differences indicated as regards susceptibility to retro- action in the above experiments. The work in Series D, using single syllables, likewise lends weight to our generalizations as stated above. PART V.. EXPERIMENTAL SECTION GENERAL SUMMARY OF THE INVESTIGATION There remains the task of bringing together the results of the various experimental series. I. The experiments involving the reconstruction test for origi- nal learning indicate very clearly that it is better to rest after the learning than to turn immediately to some other vigorous and taxing work. That is to say, learning after a rest interval can be recalled better than after a work interval. The same results have been found in the case of series of sense words as original learning. Likewise the use of single series of nonsense syllables showed the same results. However, when we used series of paired associates and tested by the method of right associates and reaction times, our results were conflicting. II. Our data indicate that the degree of similarity between the original learning and the interpolated work activity is important. Within limits, the more similar the original learning and the interpolated work the more detrimental will be the effect upon the efficiency of recall of the original material. The matter of “within limits ’’ must be stressed here. The data from the recon- struction test is quite definite in justifying the above statements. In the case of single series of nonsense syllables it was also found that the more similar the work and learning material the greater the detrimental influence on recall of the original learning. Again, the reaction times of the paired-associates (nonsense syllables) experiments failed to give clear results, although the amount reproduced indicates a greater retroaction when the work and learning are similar. Our work substantiates Robinson’s results in the matter of the importance of the degree of similarity of work and original learning. We must hasten to add certain limitations and additions to the general statements preceding : FURTHER STUDIES IN RETROACTIVE INHIBITION 57 A. When work and original learning are identical in content and method there is only reinforcement or repetition, There is no inhibition. B. As the material is made (by degrees) more and more dis- similar the reinforcing factors gradually diminish in effec- tiveness and the interfering factors become more and more pronounced. C. As the material of learning and work is made more dis- similar a point is reached where there is a maximum of interference or detrimental influence wrought upon the original learning. D. Beyond this point the curve of interference or detrimental influence goes downward, and then we can say that the more dissimilar the materials the LESS the detrimental influence. E. However, the curve of detrimental influence never reaches zero because after the work and learning are as different as can possibly be made there is still a damaging influence exerted by the work. We have found it very difficult to make out a graded series of work activities differing in degree of similarity from the original learning. III. All the data which we have accumulated indicate that the temporal position of the interpolated work is important. Work introduced immediately following the original learning is more detrimental in its influence on that original learning than work introduced after a rest interval. Reconstruction test data, single series of nonsense syllables, and paired series of nonsense syl- lables indicate the above conclusion. The little work done with sense words alone is conflicting. These results confirm the results found by Muller and Pilzecker and are opposed to the findings of Robinson, who rejects the matter of temporal position of the interpolated work as unim- portant. IV. On the basis of the data from the reconstruction experi- ments there seems evidence that as the subject becomes more and 58 ERNEST BURTON SKAGGS more fatigued in the course of the day’s series the work activity acts relatively more detrimentally upon the original learning. Tolman found that there is a more detrimental influence of work upon learning in the evening than in the morning. Our investigation of the relative effects of work activity in the morn- ing and evening, however, fail to confirm Tolman’s conclusions. When learning was as good in the evening as in the morning there was no evidence that the work interpolated was more detrimental. V. As regards the effects of practice, our data are not as positive as we would wish. As the subject becomes more and more practiced with the original learning material will there be less and less detrimental influence exerted by the work activity? Robinson has suggested this. In the case of the sense words our results actually indicate that as the subject becomes more practiced there is an increase in the detrimental action of the work material. The results from the reconstruction test are conflicting. Our data show that as the subjects become more practiced with the original learning material the learning itself is better, the subject learns more easily, and can retain longer. Certain effi- cient methods of learning become selected by the trial and error process as one becomes practiced with a given learning material. VI. Certain general statements may be made here concerning the rest and work periods. Two years of working on this inves- tigation have emphasized to us the following points: A. Trained subjects are a necessity. One cannot know about the nature of the “ rest’ interval unless the subject gives a detailed and careful report of what went on ‘“ within.” Only the subject can tell whether the work activity was “difficult or easy,” “interesting or boresome, attentive or inattentive,” etc. B. The number of “ rest” intervals which are near ideal from the standpoint of “‘ mental passivity’ and absence of all thought of the original learning is relatively few. Most rest intervals are characterized by either some “ return ”’ to consciousness of the original learning or else the subject becomes active mentally upon some line of thought. 99 66 FURTHER STUDIES IN RETROACTIVE INHIBITION 59 C. As subjects become practiced they are better able to take an indifferent and passive attitude during the rest interval. D. In most cases where there was reported some consciousness of the original learning during the rest interval it was very vague and fragmentary, often taking the form of merely desiring to recall it with a feeling that it was near at hand. E. It is important that work which will combine interest and maximum effort be secured for the work activity. We have found reasoning problems of a logical and mathematical nature the most satisfactory work material. BIBLIOGRAPHY I. Major Experimental Investigations of Retroactive Inhibition . DeCamp, J. E. A Study of Retroactive Inhibition. Psychological Mono- graph, XIX, 1915. Herne, Rosa. Uber Wiedererkennen und riickwirkende Hemmung. Zeitschrift fur Psychologie, 68, 1914, 161-236. . Mutter, G. E., and Pirzecker, A. Experimental Beitrage zur Lehre vom Gedachtniss. Zeitsch. fur Psych., Erganzungsband I, 1900, 174-198. . Ropinson, E. S. Some Factors Determining the Degree of Retroactive Inhibition. Psych. Monograph, XX VII, 1920. . ToLMAN, E. C. Retroactive Inhibition as Affected by Conditions of Learning. Psych. Monograph, XXV, 1917-1918. . Wess, L. W. Transfer of Training and Retroaction. Psych. Monograph, XXIV, 1917. . Other Experimental Studies and Discussions Bearing Upon the Problems of Retroactive Inhibition, and Inhibition in General . ANGELL and Moore. Study in Attention and Habit. Psych. Rev., III, 1896, 245-258. . Barr, J. H. The Practice Curve. Psych. Monograph, V, 1902. . Breese, B. B. On Inhibition. Psych. Monograph, III. Also, Text Book, Psychology., 1917, p. 282ff. . BrocKBANK, T. W._ Redintegration in the Albino Rat. Behavior Mono- graph, IV, 1919. . DasHteELL, J. F. Comparison of Complete vs. Alternate Methods of Learning Two Habits. Psy. Rev., 27, 1920, 112-135. . GrirFitts, C. H. Results of Some Experiments on Affection, Distribution of Associations, and Recall. Jr. Expt. Psych., III, 1920, 447-464. . Herrick, C. J. Introduction to Neurology. 1916. W. B. Saunders. . Hotmes, S. J. Studies in Animal Behavior. . Howett, W. H. Textbook of Physiology. Seventh Edition. W. B. Saunders. . Hunter, W. General Psychology, p. 304. Habit Interference in the White Rat and Human Subjects. Jr. Compar. Psych., II, 1922, 29-61. . JESINGHAUS, C. Wundt’s Psychologische Studien, 7, 1911, p. 363. 60 18. 19, 20. 7AG 22. 23. 24. 25: 26. 2h 28. a 30. J, 32, 33; 34, ERNEST BURTON SKAGGS PecHsTEIN, L. A. Whole vs. Part Methods in Motor Learning. A Com- parative Study. Psych. Monograph, XXIII, 1917. Pittspury, W. B. Fundamentals of Psychology, pp. 89f, 358f, 364f. Essentials of Psychology, pp. 213f, 221f, 215f. Pyzte, W. H. Transfer and Interference in Card Distribution. Jr. Ed. Psych., 10, 1919, 107-110. SHEPARD, J. F., and FoLtsenperGcER, H. M. Studies in Association and Inhibition. Ries). Rev., XX, 1913. é SHERRINGTON, C. S. Toteerutive Action of the Neots System. Yale Univ. Press, 1906. SmitH, W. W. Experiments on Memory and Affective Tone. British Jr. Psych., 11, 1920-21, 236-250. YERKES, R. M. Inhibition and Reinforcement of Reactions in the Frog. Jr. of Compar. Neur. and Psych., XIV, 124. Ill, Pathological Cases of Inhibition Bastian, H. C. Amnesia and Other Speech Defects. Med. Chir. Trans., LXXX, 61-86. ; Briccs, W. Verbal Amnesia Due to Shock. St. Louis Med. and Surg. Jr., 1887, LIII, 279-281. Cowtes, E. Epilepsy with Retrograde Amnesia. Am. Jr. Insanity, 1900, 593-614. Dana, C. L. The Study of a Case of Amnesia or Double Consciousness. Psy. Rev., I, 1894, 570-580. Hopkins, S. D. Amnesia, with Report of a Case. Trans. Col. State Med. Soc., 1902, 365-367. KRraFFE-EBING, R. Ueber retrograde allgemeine Amnesia. Arbeit a. d. gesamm. Geb. d. Psychiat. und Neuropath., Leipzig, 1898. Narr. Ein Fall in temporarei, totaler, theilweise Amnesia. Zeitsch. f. Hypnotismus, VI, 1897, 321-355. Paut, M. Beitrage zur Frage der retrograden Amnesie. Arch. f. Psychiat., Berlin, 1899, XXXII, 251-282. Sip1s, Boris. Psychopathological Researches. Studies in Mental Disso- ciation. New York, 1902. Srratton, G. M. Retroactive Hyperamnesia and Other Emotional Effects on Memory. Psy. Rev., 1918, 26, 474-486. (The nine preceding refer- ences given in Stratton’s bibliography.) BF21 .P96 v.34 The influence of tuition in the Princeton Theological Seminary—Speer Library UTM 1 1012 00008 5474 Aaipre Shab be bhp ody TPR eta hic tomas ; (im r tebet Pen? OPER: cetT) ; ‘ WEL PUP eres Bey. t ‘ hitet ? ' : ’ ; Ke hit r m4 ' t ef ‘ ‘ ‘ ' ‘ na) ’ " nt i i ‘ ‘ yeaa bau : { Ag ‘ bit iy : a i ‘ 4 v4 ’ Wega " ' I yen rt Behe ty t ’ i Bebe ty mari pear wry f i f I i I pieeare i ' ; i : f ‘ ' he ; / t ‘ , , vooiei eet y ; ‘ ft ¥ rear yen ux bik eee eee ij 48 wh phe 6,944 test tapi bebe bes ‘ hay e \ hat eet hs et ! H ‘ BE el EL iekt he. Rte F i , ; ‘ PEs vi i y { i j EEA UE Ee bly rer iets: f tel ' Webi tsb : ' ) i it t ' " ‘ i we ; ’ ‘ ye i ! ‘ ; Srthh baee rT it oyat § qorite’ } , i i . j i ‘ ' i ’ yea i beak ier bt Pee ‘ ' rt glee ; at ae ‘ ; i oF pet hie! iy \ Ws \ ® + ' 1 j i j ‘ shir ! ; : ‘ j f ! ap ds bi j . ; P eit ye’ ‘ (pee t r it Eva te pon Di fe er hat eed i atta sy eeu re aT heh ’ t i ears ent ta : iit j ie ' hee Ar Hea aay ; } fry , f ' Whe { LOM TAO pay ished i i j f } ‘ : iho peg ‘ ‘ ‘ 54 ke ’ ' ' ~ ' ye ’ j t ; ¥ ie ) r 44) k ; ; , t hewn de : i ‘ f ' , : i ‘ \ . f ei ‘ ) r] bigy J , j i i { ’ ; fi f tes t ' ' ‘ . ‘ ' ait nie ‘ prs ‘ ‘ Ve ‘ ; ae i fry f a ' a é ' ribet ' i 1 ‘ ery ye rg # ‘ ‘