POSITIVE TRANSFER AS A FUNCTION OF THE DEGREE OF INTER-LIST STIMULUS SIMILARITY AND INITIAL LIST LEARNING by EILEEN MARIE SHANAHAN A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS i n the Department of PSYCHOLOGY We accept t h i s thesis as conforming to the standard required from candidates for the degree of MASTER OF ARTS Members of the Department of Psychology THE UNIVERSITY OF BRITISH COLUMBIA A p r i l , 1958 POSITIVE TRANSFER AS A Ft.. /ION OF THE DEGREE OF INTER-LIST STIMULUS SIMILARITY AND INITIAL LIST LEARNING Abstract The present study was designed to test the hypothesis that positive > transfer is a function of the degree of inter-list stimulus similarity, and the degree of learning of an i n i t i a l l i s t . More specifically, the following hypotheses, derived from E. J. Gibson's theory of verbal learn-ing, were tested: 1. Positive transfer is a function of inter-list stimulus similari-ty. A decrease of inter-list stimulus similarity will result in a de-crease in the amount of positive transfer. 2. Less positive transfer will occur to a second l i s t i f practice of .ah i n i t i a l l i s t is continued after discrimination has been established among the stimulus items. The effect of the interaction between inter-list stimulus similarity and the degree of i n i t i a l l i s t learning was also assessed. Since the status of Gibson's theory did not enable the deduction of a hypothesis, the null hypothesis was tested. Sixty subjects learned an i n i t i a l l i s t of eleven stimulus forms paired with nonsense syllables of zero associative value. The subjects were required to learn each syllable so that they could spell i t when the appropriate form was presented. Learning was by the method of right as-sociates, and material was presented at the rate of two seconds per item, with a six second interval between trials. Thirty of the subjects learned this l i s t to a criterion of one perfect recitation, and the other thirty subjects learned i t to a criterion of five consecutive perfect recitations. When the criterion had been reached, the subjects were given a ten minute interval in which-to rate a series of thirty jokes. The subjects were then assigned to three groups. Each group consisted of ten subjects who had learned the i n i t i a l l i s t to a criterion of one perfect recitation, and ten who had learned i t to five consecutive perfect recitations. As a transfer task, each group received a different l i s t of paired associates, whose stimulus members were of either medium, low or zero similarity to those of the i n i t i a l l i s t . Each group learned this task to a criterion of one perfect recitation. Th8 main findings and conclusions of the study were as follows: 1 . Positive transfer is a function of the degree of inter-list stim-ulus similarity. Significantly less transfer occurs to a l i s t of zero similarity than to a l i s t of medium similarity or to one of low similari-ty. There is no significant difference between the amount of transfer to a l i s t of medium similarity and the amount of transfer to one of low simi-larity. This indicates that the relationship between positive transfer and inter-list stimulus similarity is indirect, whereas Gibson's theory indicates that the relationship should be linear. 2. Increasing the degree of i n i t i a l l i s t learning from one perfect recitation to five consecutive perfect recitations does not significantly decrease the amount of positive transfer. This was considered to be an inadequate test of Gibson's hypothesis, because the criterion of one per-fect recitation did not allow discrimination to be established among the items. 3 . There is no interaction between inter-list stimulus similarity and the degree of i n i t i a l l i s t learning. ACI3TOTO3DGEMENT The writer wishes to express her appreciation and thanks to her advisor, Dr. D. T. Kenny, for his encouragement and helpful sugges-tions. She wishes also to thank Mr. A. F. Shirran for his a s s i s -tance i n obtaining subjects for t h i s study. TABLE OF CONTENTS CHAPTER PAGE I STATEMENT OF THE PROBLEM 1 I I REVIEW OF RELATED RESEARCH 7 I I I EXPERIMENTAL MATERIALS, SUBJECTS AND PROCEDURE 14 Materials 14 Subjects 16 Procedure 18 IV THE DATA AND THEIR TREATMENT 21 V DISCUSSION OF THE RESULTS 31 VI STJMMARY AND CONCLUSIONS 56 REFERENCES 39 APPENDICES 40 A The Set of Jokes and Rating Scale 40 B Means and Standard Deviations of Cr i t e r i o n Measures for L i s t I and L i s t I I 47 TABLES TABLE I I I I I I IV V PAGE Experimental Conditions 19 Analysis of Variance of Scores on Learning of L i s t I to a C r i t e r i o n of One Perfect Recitation 22 Analysis of Variance of Number of Syllables Correct on F i r s t Recall T r i a l of L i s t I I with Three Degrees of S i m i l a r i t y and Two Degrees of L i s t I Learning 23 Comparison of F i r s t R ecall T r i a l of L i s t I I for Three Degress of S i m i l a r i t y 25 Analysis of Variance of T r i a l s to Reach C r i t e r i o n of One Perfect Recitation of L i s t I I with Three Degrees of S i m i l a r i t y and Two Degrees of L i s t I Learning 28 VI Comparison of T r i a l s to One Perfect Recitation of L i s t I I for Three Degrees of S i m i l a r i t y 29 FIGURES FIGURE PAGE Schematic Plan of a Three-Item L i s t of Paired Associates, Showing Right Excitatory Tendencies and Generalized Excitatory Tendencies 2 Stimulus Forms, Response Sy l l a b l e s , and Percent-ages of Generalization to Original Presentation of Standard Forms 17 Number of Syllables Correct on F i r s t R e call T r i a l of L i s t I I as a Function of I n t e r - L i s t Stimulus S i m i l a r i t y 26 Number of T r i a l s to One Perfect Recitation of L i s t I I as a Function of I n t e r - L i s t Stimulus S i m i l a r i t y . 30 CHAPTER I STATEMENT OF THE PROBLEM Transfer of t r a i n i n g phenomena have, as well as p r a c t i c a l s i g n i f i -cance, major implications for learning theory. E. J. Gibson»s (4) theory of verbal learning i s one of the important theories which has been pro-posed to explain and predict transfer phenomena. This theory attempts to explain transfer phenomena by applying p r i n c i p l e s and concepts from con-d i t i o n i n g theory. Gibson begins her analysis of paired associate learning by defining an excitatory tendency as "the tendency for a particular stimulus to evoke a particular response i n a capacity greater than zero" (4 p. 205). Gibson distinguishes two kinds of excitatory tendencies: r i g h t excitatory tendencies, and generalised excitatory tendencies. Right excitatory ten-dencies, set up between the stimulus and response members of a pair, lead to correct responses. Generalized excitatory tendencies, set up between sti m u l i and responses other than members of a pair, are the re s u l t of generalization, or a lack of discrimination, between two stimulus itsms i n a l i s t . The strength of any generalized excitatory tendency i s a func-t i o n of the degree of generalization between the stimulus members involved. Since a l l s t i m u l i within a l i s t w i l l not necessarily generalize to the same degree, generalizing tendencies may vary i n strength. Figure I, a schematic plan of a l i s t of paired associates, indicates the two kinds of tendencies. The effect of generalized excitatory tendencies w i l l be determined by the relationship between the response members of the gener-a l i z i n g s t i m u l i . 2 Figure 1. Schematic plan of a three-item l i s t of paired associates, showing r i g h t excitatory tendencies ( and generalized excitatory tendencies ( ^ ). ( After Gibson, 4, p. 198. ) 3 I f a ri g h t excitatory tendency and a generalized one are evoked by the same stimulus, the resultant of the two tendencies w i l l be stronger than either tendency alone. I f the responses to two generalizing s t i m u l i are the same, the right excitatory tendency and the generalized excitatory tendency w i l l coincide. Thus, i f S a and S D generalize and R a and Rfc a r e the same, S a —:>R a and S a — R D w i l l coincide. The resultant of these two tendencies w i l l be stronger than either Sa—^>Ra o r Sa > Rb alone. Since the strength of Sa—7" Rb i s a function of the degree of the general-i z a t i o n between S a and Sb , learning w i l l be easier i n proportion to the degree of generalization between S a and S D. I f the responses to two generalizing s t i m u l i are dif f e r e n t , the r i g h t excitatory tendency w i l l be blocked by the generalized excitatory tendency, i n proportion to the strength of ths l a t t e r . Thus, i f Sa and Sb general-ize and R a and Rb are diff e r e n t , S a ^R a w i l l be blocked by S a — ^ Rb, i n proportion to the strength of S a—^- Rfc. The greater the degree of general-i z a t i o n between Sa and Sb, the stronger w i l l be Sa—^Rb» and the greater the blocking. Generalizing tendencies must be weakened i n order that the correct response can occur. These can be weakened only by rei n f o r c i n g correct responses. Gibson defines reinforcement as "a process which oc-curs during verbal learning when a subject sees a response as he had an-tici p a t e d i t , and thinks 'That's r i g h t * M (4, p. 205). The amount of re-inforcement required to weaken any generalizing tendency w i l l be a func-t i o n of the strength of that particular generalizing tendency. When a l i s t of paired associates i s presented for learning, there i s an i n i t i a l increase i n the tendency for stimulus members to generalize with each other. Generalization reaches a peak early i n learning. As 4 practice i s continued, generalization progressively decreases u n t i l a l l the s t i m u l i are discriminated from one another. Thereafter, these s t i m u l i w i l l tend to generalize less with new stimulus items, the decrease i n generalization "being proportional to the amount of d i f f e r e n t i a l r e i n -forcement given. So f a r this discussion has been concerned only with generalization between the members of one l i s t . But Gibson states that i f stimulus mem-bers generalize with each other when presented i n the form of one l i s t , they w i l l do so, with the same re l a t i v e degree of generalization, when presented i n the form of two l i s t s (4, p. 207). Gibson states that positive transfer w i l l occur when the sti m u l i of two l i s t s generalize, i f the nature of the discrimination established i n the f i r s t l i s t i s ben e f i c i a l to the second l i s t . Thus, i f each stimu-lus member of one l i s t has a generalizing member with the same response i n the second l i s t , learning w i l l be f a c i l i t a t e d . The greater the degree of generalization between the stimulus members of the two l i s t s , the great-er w i l l be the positive transfer. One of the conditions a f f e c t i n g generalization i s s i m i l a r i t y , which Gibson defines as, "a condition e x i s t i n g between stimulus members which causes them to generalize" (4, p. 208). Therefore, the greater the simi-l a r i t y between two s t i m u l i , the greater w i l l be t h e i r tendency to generalize with each other. I t would be predicted that i f the stimulus members of two l i s t s are s i m i l a r , and i f s i m i l a r members have the same response, then the greater the s i m i l a r i t y between the stimulus members of the two l i s t s , the greater w i l l be the positive transfer. The present study i s an attempt to test this hypothesis. Three groups 5 of subjects learned the same i n i t i a l l i s t of paired associates. Then each group was given a different transfer task. One group learned a second li s t of paired associates whose stimulus members were of medium similarity to those of the i n i t i a l l i s t . The second group learned one whose stimulus members were of a low degree of similarity. The third group learned a l i s t whose stimulus members were of zero similarity. It was predicted that there would be less positive transfer to a l i s t of low or zero similarity than to one of medium similarity, and less to a l i s t of zero than to one of low inter-list similarity. Secondly, this study tests, with respect to positive transfer, the hypothesis that i f practice of a l i s t is continued after the stimulus members of the l i s t have been discriminated from each other, they will generalize less with new stimuli. According to Gibson, less positive trans-fer should occur to a second l i s t i f practice of an i n i t i a l l i s t is con-tinued after discrimination has been established among the stimulus items. Although Gibson has not operationally defined when discrimination will be established, it appears to be at the point when right responses will be evoked by a l l the stimulus members of the l i s t . Therefore, for the purpose of this study, when a subject has learned the l i s t to one perfect recitation the point of discrimination has been reached. Consequently, when a subject has learned a l i s t to five consecutive perfect recitations, the point of discrimination has been passed. Two groups of subjects learn-ed the same in i t i a l l i s t of paired associates. One group learned the l i s t to one perfect recitation, the other learned it to five consecutive -perfect recitations. The same transfer task was then given to both groups, who learned this l i s t to one perfect recitation. It was predicted that less 6 positive transfer would occur when the l i s t was learned to a criterion of five consecutive perfect recitations than i f i t were learned to one perfect recitation. This experimental design also makes i t possible to assess the signi-ficance of interaction betvjeen inter-list stimulus similarity and the degree of learning of the in i t i a l l i s t . Since no prediction could be made from Gibson's theory, a null hypothesis was tested. CHAPTER I I REVIEW OF RELATED RESEARCH Int e r - L i s t S i m i l a r i t y Studies The f i r s t experimental treatment of the hypothesis that positive transfer i s a function of i n t e r - l i s t stimulus s i m i l a r i t y was reported by Yum (11) i n a series of three studies. In the f i r s t study, subjects were required to learn a l i s t of hyphenated nonsense syllab l e s t i m u l i (e. g. REB-QIM) paired with f o u r - l e t t e r word responses (a. g. WOLF). As a trans-fer task, a second l i s t , i n which stimulus s i m i l a r i t y was varied by chang-ing one or more l e t t e r s of the hyphenated syllables and varying the pos i t i o n of the changed l e t t e r s , was presented. The degree of stimulus s i m i l a r i t y , defined as the number of l e t t e r s changed, did not s i g n i f i c a n t l y a l t e r the amount of positive transfer. However, the locus or position of change proved s i g n i f i c a n t . Changing the f i r s t or the middle l e t t e r s of either s y l l a b l e reduced transfer s i g n i f i c a n t l y , but changing the l a s t l e t t e r of either s y l l a b l e did not. Furthermore, changing the f i r s t l e t t e r of either syllable resulted i n s i g n i f i c a n t l y less transfer than changing the middle l e t t e r of either s y l l a b l e . In the second experiment of the series, two groups of subjects learned a l i s t of paired associates with meaningful word stimulus and response members. As a transfer task, one group received a l i s t with stimulus members quite similar i n meaning to those of the i n i -t i a l l i s t and the other group received one with stimulus members moderately similar ( s i m i l a r i t y of meaning had previously been determined by judges» rat i n g s ) . There was s i g n i f i c a n t l y more transfer when s i m i l a r i t y was great-est. In the t h i r d study, four groups- of subjects learned a l i s t consisting 8 of v i s u a l stimulus patterns paired with meaningful word responses. As a transfer task, each group received one of four l i s t s , the stimulus members of each l i s t judged to be of a different degree of s i m i l a r i t y to those of the i n i t i a l l i s t . The results indicated that positive transfer was a func-t i o n of i n t e r - l i s t stimulus s i m i l a r i t y . In his study, McKinney (9) used four geometric figures as s t i m u l i , to which subjects responded with various l e t t e r s of the alphabet. The amount of transfer, when stimulus figures were altered 10%, 20% or 30%, was tested. For example, one of the figures was a four centimetre l i n e perpendicular to an eight centimetre l i n e at i t s center, producing a figure which looked l i k e a cross with one arm missing. A 10% a l t e r a t i o n of t h i s figure would be one with 10% taken o f f any arm, or 3 l / 3 % taken off each arm simultan-eously. Alterations of 20% and 30% were made i n a similar manner. Per-centage of transfer was found to decrease as a function of the degree of a l t e r a t i o n , i f the locus of a l t e r a t i o n remained constant. However, the percentage of transfer was not proportional to the percentage of a l t e r a -t i o n . More transfer was obtained when the st i m u l i were altered symmetri-c a l l y ( i . e., when a l l the arms of the figure were reduced simultaneously) than when the s t i m u l i were altered asymmetrically ( i . e., when only one or two arms of the figure were reduced). McKLnney's explanation was that the meaning or quality of a stimulus i s changed when i t i s altered asy-mmetrically: since a stimulus i s a function not only of i t s mass or quan-t i t y , but also of the manner i n which that quantity i s dis t r i b u t e d . Hamilton (6) studied both positive transfer and retroactive f a c i l i t a -t i o n as a function of i n t e r - l i s t stimulus s i m i l a r i t y . The stimulus figures used i n this study were those standardized by Gibson (5). These materials 9 consisted of thirteen forms (the standard l i s t ) , and three sets of varia-tions of the standard l i s t . In the first l i s t the figures were of medium similarity; in the second, of low similarity; and in the third, of zero similarity. The thirteen figures of the standard l i s t were paired with nonsense syllables, and presented for learning to five groups of subjects. When a criterion of 8/l3 correct responses had been reached, learning was discontinued. A four-minute rest period, during which the subjects read Life or The Mew Yorker, or talked to the experimenter, followed for four of the groups. A second task was introduced: the first group relearned the standard l i s t , the second received the medium similarity l i s t , the third received the low similarity l i s t , and the fourth received the zero similarity l i s t . Each group learned the second l i s t to a criterion of 8/13 correct responses in approximately sixteen minutas. The f i f t h group had a twenty minute rest period, during which they read Life or The New Yorker, or talked to the experimenter. Then a l l groups rslearned the standard l i s t to a criterion of one perfect recitation. A gradient of positive transfer as a function of inter-list stimulus similarity was ob-tained, but only the difference between the group receiving the medium similarity l i s t and the zero similarity l i s t was significant. The same gradient was obtained using retroaction measures, and again only the dif-ference between the group receiving the medium similarity l i s t and the group receiving the zero similarity l i s t was significant. Bugelski and Gadwallader (2) also used Gibson's (5) standardized stimulus figures in their study of transfer and retroaction effects. The thirteen figures of the standard l i s t were paired with meaningful words, and presented for learning to four groups of subjects. Tha materials were 10 presented on cards, and an item was dropped from the pack when i t had been correctly anticipated on two successive t r i a l s . When a l l the items had been learned to t h i s c r i t e r i o n , a l l the groups were given a two min-ute rest period. A second task was then introduced: the f i r s t group re-ceived the medium s i m i l a r i t y l i s t , the second group received the low sim-i l a r i t y l i s t , and the t h i r d received the zero s i m i l a r i t y l i s t . The c r i -t e r i o n of learning for t h i s l i s t was the same as for the i n i t i a l l i s t . The fourth group read The New Yorker for eight minutes. After a two min-ute rest period a l l groups were tested for r e c a l l of the standard l i s t . P ositive transfer increased as a function of i n t e r - l i s t stimulus s i m i l a r -i t y , but the only s i g n i f i c a n t difference was that between the group re-ceiving the medium s i m i l a r i t y l i s t and the one receiving the zero s i m i l -a r i t y l i s t . Retroactive f a c i l i t a t i o n increased as a function of i n t e r -l i s t stimulus s i m i l a r i t y . Studies of the Degree of I n i t i a l Task Learning Atwater (1) used a verbal learning s i t u a t i o n to study positive trans-fer as a function of the degree of i n i t i a l task learning. Four groups of subjects learned a l i s t of ten paired associates, consisting of three-l e t t e r words as stimulus and response members. The degrees of learning for the four groups ware: no learning, six correct responses, one perfect r e c i t a t i o n plus five t r i a l s , and one perfect r e c i t a t i o n plus f i f t e e n t r i a l s . The four groups were given the same transfer task, that of learn-ing the responses of the i n i t i a l l i s t to new s t i m u l i . A l l groups learned 11 t h i s l i s t to a c r i t e r i o n of one perfect r e c i t a t i o n plus f i v e t r i a l s . P ositive transfer was a function of the degree of i n i t i a l task learning, and a l l differences i n the amount of transfer among the four degrees of learning were s i g n i f i c a n t . In his study, Mandler (V) used a switchboard apparatus which had s i x switches arranged i n a hexagon. Five groups of subjects responded to each of four l e t t e r s of the alphabet by operating the correct sequence of three switches on t h i s switchboard. The groups were given 0, 10, 30, 50, or 100 errorless t r i a l s . After the c r i t e r i o n had been reached, there was a three minute rest period. The same transfer task, learning the responses of the i n i t i a l task to new s t i m u l i , was then given to a l l groups. A l l groups learned t h i s task to a c r i t e r i o n of two successive, errorless repetitions or twenty t r i a l s , whichever occured l a s t . P o sitive transfer was a function of the degree of i n i t i a l task learning, but only the d i f -ference between the group which had no learning and the one given 100 error-less t r i a l s was s i g n i f i c a n t . In an attempt to determine the generality of Handler's (7) findings for motor behavior, Mandler and Heinemann (8) have conducted a comparable study using verbal materials. In t h i s study the procedure was the same as for Mandler's (7) study, except that the materials used were single integer numbers as s t i m u l i , paired with three-place consonant nonsense sy l l a b l e s . P ositive transfer increased as a function of the degree of i n i t i a l task learning. There was no s i g n i f i c a n t difference i n the amount of positive transfer with 0, 10, or 30 errorless t r i a l s , but with 50 error-less t r i a l s there was a s i g n i f i c a n t increase of positive transfer. With 12 100 errorless t r i a l s there was s i g n i f i c a n t l y more transfer than with 50 t r i a l s . A much greater increment i n the number of t r i a l s was required to produce a s i g n i f i c a n t increase i n the amount of positive transfer when motor learning was involved than when verbal materials were used. Col-lege students were the subjects for both studies, and the differences between the results of the two studies were explained i n terms of the subjects used. College students are more adept at recombining units of verbal behavior than those of motor tasks. Studies Investigating Inter-Task Stimulus S i m i l a r i t y and Degree of I n i t i a l Task Learning and Interaction Between Them The only available study of positive transfer as a function of both inter-task stimulus s i m i l a r i t y and the degree of i n i t i a l task learning i s that of Duncan (3). He used an apparatus consisting of s i x s l o t s ar-ranged r a d i a l l y on a panel. While holding a lever steady with h i s l e f t hand, each subject responded to each of six colored l i g h t s t i m u l i by mov-ing, with his ri g h t hand, a driver into the correct s l o t . The subjects were given 10, 40, 80 or 180 t r i a l s of t h i s task. Then the subjects were given one of three transfer tasks, each of a di f f e r e n t degree of stimulus s i m i l a r i t y . In the most similar task, four l i g h t - s l o t combina-tions remained the same as for the f i r s t task; i n a l e s s similar task, only two l i g h t - s l o t combinations remained the same; i n the least simi-l a r task, a l l the l i g h t s were paired with different s l o t s . Transfer was positive for a l l groups, and was a function of both the degree of i n t e r -task stimulus s i m i l a r i t y and the degree of i n i t i a l task learning. He 13 found no interaction between inter-task stimulus s i m i l a r i t y and the de-gree of i n i t i a l task learning. Summary of Related Research Available data lend support to the hypothesis that positive trans-fer i s a function of inter-task stimulus s i m i l a r i t y . The hypothesis that positive transfer decreases with a high degree of learning of the i n i t i a l task i s not supported. Positive transfer has been found to i n -crease with high degrees of i n i t i a l task learning. No available study indicates interaction between inter-task stimulus s i m i l a r i t y and the de-gree of i n i t i a l task learning. CHAPTER I I I EXPERIMEETAL MATERIALS, SUBJECTS AND PROCEDURE Experimental Materials The stimulus materials used i n t h i s study are those which were stan-dardized by E. J. Gibson (5) for use i n her study of retroactive i n h i b i -t i o n as a function of the degree of generalization between tasks. Gib-son drew thirteen standard forms and a number of variations of each of them. These forms were given to a group of ten judges, who were asked to select, for each standard form, the va r i a t i o n which was most similar to the standard, one which was less s i m i l a r , and one which was d i s s i m i l a r . This procedure gave four sets of forms: thirteen standard forms, thirteen variations of medium s i m i l a r i t y , thirteen variations of low s i m i l a r i t y , and thirteen d i s s i m i l a r variations. A l i s t composed of the thirteen standard forms, each paired with a different nonsense s y l l a b l e , was given to a group of subjects to learn by the paired associates method. Twenty-four hours l a t e r , the subjects were given one of several r e c a l l l i s t s . Each of these l i s t s was composed of four standard forms, three variations of medium s i m i l a r i t y , three variations of low s i m i l a r i t y , and three variations of no s i m i l a r i t y , so that, i n each l i s t , each standard form was represented by the standard form or by a va r i a t i o n . Gibson calculated an objective measure of gen-e r a l i z a t i o n for each standard fonn and v a r i a t i o n by determining the per-centage of subjects who responded to each form by giving the response with which the standard form was paired i n the f i r s t l i s t . In only two 15 instances was there disagreement between the objective and subjective ratings. The judges had rated one va r i a t i o n as being of low s i m i l a r i t y and another as being of zero s i m i l a r i t y . Objective measures showed a reversal of these ratings. When the l i s t s were arranged i n agreement with the objective ratings, the res u l t was four l i s t s of thirteen figures each: the standard l i s t , with an average of 84.5$ generalization to thei r o r i g i n a l presentation; the medium s i m i l a r i t y l i s t , with an average of 41.1% generalization to the o r i g i n a l presentation of the standard forms; the low s i m i l a r i t y l i s t , with an average of 9.7$ generalization to the o r i g i n a l presentation of the standard forms; and the zero s i m i l a r i t y l i s t , which generalized 0% with the o r i g i n a l presentation of the standard forms. In the present study only eleven of the thirteen forms are used. Yariations of two of the standard forms i n both the low and the zero sim-i l a r i t y l i s t s yielded measures of zero generalization. Otherwise, Gibson's four l i s t s were used without further a l t e r a t i o n . To each of the eleven standard forms a nonsense s y l l a b l e of zero associative value, drawn from Glaze's (10) calibrated l i s t s of rt'onsense s y l l a b l e s , was assigned. The same nonsense s y l l a b l e was assigned to a form and a l l i t s variations. To reduce response generalization, two conditions were imposed upon the selection of the syll a b l e s : no two syllables may begin with the same l e t t e r , and no two syl l a b l e s may end with the same l e t t e r . Figure 2 shows the eleven forms with t h e i r respective variations, nonsense s y l l a b l e names, and the percentages of generalization as determined by Gibson. Twenty-four random orders of the standard forms were used i n the 16 o r i g i n a l learning and r e c a l l situations.' Bach l i s t v/as constructed by-drawing, with India ink on a sheet of drawing paper, the thirteen stand-ard forms, one below the other. Each stimulus figure was -J" x The respective nonsense s y l l a b l e names were typed i n c a p i t a l l e t t e r s to the ri g h t of each stimulus form. The r e c a l l orders were constructed i n the same way, except that blank spaces were substituted for response names. Since subjects were to learn the standard l i s t by the method of righ t associates, the material was mounted on a memory drum so that a learning order was always followed by a r e c a l l order. A blank space was l e f t bet-ween each order so that no material was v i s i b l e when the memory drum was stopped between t r i a l s . The memory drum was regulated to allow an ex-posure time of two seconds per item. Material for the transfer task was prepared and mounted i n the same manner. A series of t h i r t y jokes, which the subjects were instructed to rate on a five-point humour rating scale, was prepared for use as an i n t e r p o l -ated a c t i v i t y during the ten minutes after the completion of the learning of the standard l i s t . This a c t i v i t y was introduced to enable the experi-menter to arrange the materials for the transfer task, and to reduce the subject's fatigue. This material w i l l be found i n Appendix A. Subjects The si x t y subjects were volunteers from a class i n Introductory Psy-chology and from a class i n Psychological Testing at the University of B r i t i s h Columbia. Subjects, i f they were unable to learn the i n i t i a l l i s t i n the a l l o t t e d time of-one hour, were eliminated. 17 STANDARD MEDIUM LOW ZERO % of Forms general-i z a t i o n % of Forms general-i z a t i o n % of % of Forms general- Forms general- RB-i z a t i o n i z a t i o n SPONSE / 93% 76% 96% / 84% 72% 25% 11% 7% 0% YUZ TOY GOQ, 93% 28% 8% 0% MEF p 90% P . 83% 32% 0% ZIL 88% CO 21% 0 17% 0% WUH 74% 21% A 4% 0% JIC 88% 11% 10% 0% QUS 67%