WWW \\W W Lx 3 1293 01005480 __-..———‘ ___..-——- __,_.———-— / AUG/*1 1 2000 ‘ I ABSTRACT THE EFFECT OF ALTERNATING SOCIAL APPROVAL COMMENTS AND TANGIBLE REWARDS ON TASK PERFORMANCE OF KINDERGARTEN CHILDREN By Richard Allan Brown Treatment programs for children based on operant condi- tioning principles often utilize social approval comments and tangible rewards like candy and toys as positive rein- forcement. This study focused on the following questions: 1) Are tangible rewards generally more effective than social approval comments in influencing performance of kindergarten children?, 2) Can interaction effects be caused by eXposure to two different magnitudes of reinforce- ment such as tangible rewards (high magnitude) and social approval comments (low magnitude)?, 3) Does superior perfor~ mance that has been maintained by tangible reinforcement remain superior in a non-reinforcement situation when com- pared with performance that has previously been maintained by social reinforcement?, and 4) Can direct teacher ratings on cooperation, persistence, reSponse to approval comments, and reSponse to tangible rewards predict children's perfor- mance in a task situation under conditions of social approval, or tangible reward? The task was key pressing. Twenty kindergarten children ‘were assigned to each of three experimental groups under social approval comments from an adult, tangible reinforce- ment, or alternated social and tangible reinforcement. To Brown test for confounding sex effects an equal number of boys and girls was included in each group. To test for confounding effects due to satiation half of each group was run under high or low density schedules of reinforcement. Trials con- sisted of 20 second presentations of the key. There were 4 trials following initial instructions, then 20 trials under reinforcement, and 8 trials under no reinforcement. There was no evidence that tangible reinforcement was better than social reinforcement except when they were alter- nated. Rerformance under the tangible condition of the alternated groupincreased more than performance under any of the other conditions. Alternation with social reinforce- ment increased the influence of the tangible reinforcement. Under the non-reinforcement period performance of the group with a tangible reinforcement history dropped below the groups with social and alternated reinforcement histories. This was discussed in terms of Amsel's frustration hypothesis and the notion that the tangible reinforcement situation was more unlike non-reinforcement than the other two reinforcement conditions. Girls were more reSponsive to reinforcement than boys and there were no confounding interactions with other variables. In the non-reinforcement period there was an interaction between density of reinforce- ment and trial blocks but there were no confounding inter- actions with other variables. Correlations between rating ~scales filled out by teachers were of no value in predicting children's performance in the eXperimental situation. THE EFFECT OF ALTERNATING SOCIAL APPROVAL COMMENTS AND TANGIBLE REWARDS ON TASK PERFORMANCE OF KINDERGARTEN CHILDREN BY Richard Allan Brown A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1969 C. (0' '2/ 7 (I, 2 3' 1/5.... ACKNO E-IL ED GEMEN TS I would like to thank my chairman, Dozier Thornton, for his patience in the development of this thesis. I would like to thank Lucy Ferguson and Terry Allen for their helpful suggestions in the design of the study and final revisions of the manuscript. And I would like to thank Bill Kell for giving me his attention in some very helpful ways related more to the motivational problems of writing this thesis than the specific content. ii ll [[ [Ill 1 l l I I I I i in I I llllll LIST OF TABLES. LIST OF FIGURES INTRODUCTION. TABLE OF STATEMENT OF THE PROBLEM. METHOD. . RESULTS . DISCUSSION. BIBLIOGRAPHY. APPENDIX. CONTENTS iii iv 10 12 22 39 49 54 LIST OF TABLES Analysis of Variance, Alternated Social Reinforcement. . . Analysis of Variance, Alternated Tangible Reinforcement. . . Analysis of Variance, Non-Reinforcement. . Intercorrelations of Teacher Ratings . . . Correlations between Teacher Ratings and Difference Scores. iv 26 30 33 36 37 LIST OF FIGURES Apparatus During Rest Periods. . . . . . . . . Apparatus During Base and Non-Reinforcement Periods O O O O C O O O O O O O O O O O O O 0 Apparatus During Tangible Reinforcement PeriOdS. O O O O O O O O O O O O O O O O O 0 Apparatus During Social Reinforcement Periods. Comparison of Groups under Different Kinds of Reinforcement in Conditions of Reinforcement and Non-Reinforcement. . . . . . . . . . . . Performance of Two Subjects Alternated Between Tangible and Social Reinforcement. . . . . . Performance of Boys and Girls During the Reinforcement Period for All Groups. . . . . Performance of High and Low Density Reinforce- ment Groups during the Non-Reinforcement PeriOd O O O O O I O O O O O O O O O O O O O 15 15 l6 16 24 27 29 34 INTRODUCTION During the past few years the author has been asso- ciated with a day school project for emotionally disturbed children conducted by the Psychology Department at the In- stitute for Juvenile Research (Perce, 1968; Brown, Pace & Becker, In Press). In this project there was heavy empha- sis on Operant conditioning principles, particularly the use of tangible reward systems in the classroom which were styled after those of wolf, Giles & Hall (1968) and Birnbrauer, wolf, Kidder & Tague (1965). The children in the Institute for Juvenile Research (IJR) program could earn in excess of 2000 points a week for academic work and good behavior in the classroom. These points could be used to "buy" toys at the end of each schoolday as well as for planned parties and trips. In addition, candy and treats were distributed during the class periods to reward specific behaviors. Within a few months these children were behaving very nicely in the class- room. However, outside the classroom they were often still unmanageable when treated with ordinary social conventions. As these children passed through the doorway of the classroom a Dr. Jekyll - Mr. Hyde phenomenon was observed. Passing over the threshhold from the classroom to the outside the children became "wild" but in returning over the same 2 threshhold they became very quiet, polite, and generally well-behaved. At a time when tangible reward systems are more frequently included as part of treatment programs, ex- perience with our project has led us to ask several questions about the application of these kinds of systems: 1) Are tangible rewards more effective than social approval in shaping and maintaining desired reSponses? It was our impression that many of the desired behaviors tangibly rewarded in our classroom situation had occurred previously in natural social interaction. Conventional re- wards from adults in natural social interaction are usually little more than approval, often expressed verbally by state- ments like "Good", "Fine", "M-h'ml", "You're doing a nice job!", etc. The potency of these kinds of statements as re- wards has been demonstrated by a number of experiments (Ger- witz & Baer, l958a, 1958b; Stevenson & Odom, 1961; Green & Zigler, 1962; Hill & Stevenson, 1964; Patterson & Hinsey, 1964; and Stevenson & Fahel, 1965). Approval comments not only influence children's behavior but are more effective after conditions of social "deprivation" and less effective after social "satiation". The effectiveness of contingent attention from adults has been demonstrated in a variety of treatment situations. The form of this attention has varied from approval comments to smiles and physical contact. The .usual procedure is to attend to desired behaviors which are often chosen so that they will be incompatible with inappro- priate behavior. Inappropriate behaviors are usually ignored as much as possible. Variations of this procedure have been effective for treating children's problems in nur- sery school situations (Harris, Johnston, Kelly & wolf, 1964; Hart, Allen, Buell, Harris & wolf, 1964; Allen, Hart, Buell, Harris & Wolf, 1964; and Brown and Elliott, 1965), natural classroom situations (Becker, Madsen, Arnold & Tho- mas, 1967; Madsen, Becker & Thomas, 1968; and Thomas, Becker & Armstrong, 1968), and simulated home situations (Wahler, Winkel, Peterson & Morrison, 1965; Johnson & Brown, In Press). When a tangible reward system is employed as part of a treatment program for children the rationale is that the oc- currence of desired reSponses can be increased more quickly and maintained at higher rates than under contingent social approval alone. Tangible reward systems ranging from highly elaborate token systems to simple direct rewards such as food and trinkets have been utilized in institutional and school programs (Ayllon & Azrin, 1968; Wolf 35 g;., 1968; Birnbrauer'gt‘a;., 1965; O'Leary & Becker, 1969) and in in- dividualized treatment programs, particularly with autistic and schiZOphrenic children (Brown,g£,§;., In Press; Ferster & DeMyer, 1961; Wolf, Risely & Mees, 1964; Lovaas, Berher- ich, Perloff & Schaffer, 1966). In cases where children are not responsive to adult approval or in cases where adult ap- proval is aversive, tangible reinforcement is clearly more Jeffective during the initial stages of treatment (Brown g£,§1., In Press). There is little empirical evidence, however, that 4 tangible rewards are generally more effective than social approval, particularly where children are not systematically deprived of food, toys, etc. Examination of treatment pro- cedures where tangible reinforcement is utilized often re- veals a strong component of social reinforcement in the form of instructions and distribution of the rewards by adults. Because tangible rewards may have value both as tangible re- wards and as tokens of social approval they are probably more potent than social approval alone. 2) Can interaction effects be caused by eXposure to two different magnitudes of reinforcement such as tangible rewards (high magnitude) and social approval comments (low magnitude)? In the IJR project it appeared that the pervasive use of tangible rewards in class may have made the more natural rewards available outside the classroom, primarily social ap- proval, less effective. The Dr. Jekyll - Mr. Hyde phenomenon may be analogous to the results of eXperiments demonstrating contrast effects. Dunham (1968) has recently reviewed these experiments and has labelled the results of "within - subjects" methodology "behavioral contrast" and the results of "between - subjects" methodology "incentive contrast". Behavioral con- trast is a shift in performance from a situation under one or more magnitudes of reinforcement to a situation in which some new'magnitude of reinforcement is introduced, where the shift tin performance under the old level or levels of reinforcement is in an Opposite direction from the new magnitude of 5 reinforcement. Incentive contrast is a shift in performance caused by an upshift or downshift in magnitude of reinforce- ment where the increase or decrease in performance exceeds the eXpected level of performance indicated by control groups under only the high or low magnitudes of reinforcement. In Dunham's (1968) critical review of the contrast phe- nomena he finds some support for positive contrast effects, strong support for negative contrast effects, and in some experiments in which high and low magnitudes of reward have been alternated there is some evidence for a biphasic pro- cess in which the negative contrast effect is preceded by facilitation of performance. Experiments employing alterna- tion procedures are most analogous to the IJR project in which the classroom situation (high magnitude rewards) was alternated with the natural environment (low magnitude re- wards) outside the class. Two of these experiments are sum- marized below: Peiper & Marx (1963) magazine trained 3 groups of rats in a Skinner Box. The groups were trained on 4%, 11.3%, or 32% concentration of sucrose solution. Magazine training was then alternated with bar press training at 11.3% sucrose for all groups. The downshifted group first reSponded higher and then lower than the unshifted control resulting in a sta- ble negative contrast effect. The upshifted group first reaponded lower and then higher than the non-shifted group .resulting in a stable positive contrast effect. Bower (1961) trained rats in two distinctly different alleys. The experimental group was alternately trained with 6 a large magnitude of reward (8+) in one alley and a small magnitude of reward in the other alley (3-). Control groups were run only under the large or small magnitude of reward. When the 5- condition of the eXperimental group was compared with controls run only under the small magnitude condition the experimental group was at first significantly faster but in later trials significantly slower in their performance. A similar experimental procedure of Glass & Ison (1966) also resulted in a facilitation effect followed by a negative contrast effect. If conditions existed in our treatment project that were analogous to these contrast experiments there may have been several possible interaction effects. The classroom situation with a high density tangible reinforcement system was alternated with the natural environment outside of the classroom where contingent adult approval was the primary reinforcement. Alternation with social approval could have accentuated the effect of the tangible reinforcement creating a Dr. Jekyll effect (positive contrast). Alternation with tangible reinforcement could have decreased the effect of social rein- forcement creating a Mr. Hyde effect (negative contrast). Over a series of alternations with tangible reinforcement the effect of social approval could have first been facili- tated and then decreased (biphasic effect). 3) Does superior performance that has been maintained by tangible reinforcement remain superior in a non-reinforce- ment situation when compared with performance that has 7 previously been maintained by social reinforcement? An important question for the IJR project was whether the apprOpriate classroom behavior maintained by high den- ~4 sity tangible reinforcement would transfer back to the local school situations where there was no tangible rein- forcement and where the teachers often provide minimal social reinforcement. Hulse (1958), Armus (1959), and Wag- . ner (1961) have found faster extinction following larger . Aw magnitudes of reward. This suggests that although tangible reinforcement may originally maintain higher levels of res- ~' "'9. ponse, response in a non-reinforced situation may be less than if the response were originally maintained at a lower , level with social reinforcement. Amsel (1958, 1962) makes a strong argument toward an active role for non-reward. He believes non-reinforcement can produce frustration reSponses which are incompatible with the previously rewarded responses. Many sequences of interaction observed outside the IJR classroom situation seemed to confirm his views. They were usually variations on the following theme: Adult makes a request of child. Child replies, "What do I get?" Adult says, "Nothing" and repeats request. Child is observed to do various things like scream, cry, break things, sit like a stone, use abusive language, stomp feet, make fierce faces, make sad faces, f10p around on the floor, bang head, strike adult, ‘suck thumb, run away, etc. Is it possible that a high density tangible reinforcement 8 system such as we had in the classroom.could contribute to frustration and poor performance in situations where desired responses are not reinforced with tangible rewards? " 4) Can direct teacher ratings on cooperation with adult instructions, persistence, reSponse to adult approval comments, and response to tangible rewards predict child- ren's performance in a task situation under conditions of ‘4 social approval, or tangible reward? A In the IJR project there were many individual differ- ences between the children including their response to var- ‘; I ECWI‘. ‘ ious kinds of reinforcement. There was a large variation in response to the tangible rewards as well as to social appro- I val comments from adults. It would be useful in setting up individual therapy and remediation programs if teachers could accurately predict how a child would respond to various kinds of reinforcement. Some attempts have been made to relate scales and rat- ings to performance in social approval situations but the results have been less than impressive. Marlowe (1962) indi- cated that need for social approval facilitates verbal con- ditioning, however Spielberger, Berger & Howard (1963) found no relationship between Crown - Marlowe scores on need for approval and rate of verbal conditioning. In another study Costello (1967) found no relationship between the effect of social approval comments on task performance of pre-school Ichildren and teacher ratings of the children on a "social competence" scale by Kohn & Silverman (1966). Many items 9 on this scale are general and it may be that more specific ratings by teachers would be more fruitful in predicting the performance of individual children. 2 :,W.‘ Jr _ STATEIENT OF THE PROBLEM This study is intended to be an eXperimental analogue 3, designed to answer some of the major questions raised above. Answers to these questions are critical in the develOpment of treatment programs for children which include the use of tangible reinforcement systems. Given one group of children .1 reinforced with tangible rewards (T), another group with ' AL social approval comments from an adult (8), and another group alternately reinforced with tangible rewards and so- cial approval comments (A): 1) Will T perform better than S? " 2) Will there be positive contrast, negative contrast, and/or biphasic effects in A? 3) Will T, S, and A differ in a non-reinforcement situation? 4) Can a child's reSponse to social approval and tan- gible reinforcement be predicted by his teacher? Specific Hypotheses 1) Performance may be higher under tangible reinforce- ment than under social approval comments (TV:>S). 2) Where tangible reinforcement is alternated with social approval comments a) performance under the tangible alternation may first be lower (T:>’AT) and then higher 1(AT:>'T) than the performance of the group under tangible reinforcement alone, and b) performance under the social alternations may first be higher (A§=> S) and then lower 10 11 (S:=’AS) than the performance of the group under social approval comments alone. 3) In the non-reinforcement situation immediately following the reinforcement condition performance of the group with a tangible history may not be significantly higher than the group with a history of social reinforce- ment. It is possible that there may be more of a frustra- tion effect associated with a history of tangible rewards. If this is the case, performance of group T may drop below group S in the non-reinforcement situation (SZ>’T), and performance of group A may fall between the other two groups (S>A> T). 4) Teacher ratings may predict children's performance in reSponse to tangible rewards and social approval comments. 1" law‘VJhl'JSSl 1mm'm-a 1' "' s“ g Contrast effects were tested by havinglg's Operate a telegraph key under conditions of tangible, social, or alternated tangible and social reinforcement. The rein- forcement period was followed by a period of key presen- tation without reinforcement. Because of the possibility that there might be confounding interaction effects due to sex and satiation, sex and density of reinforcement were included as variables in the eXperimental design. Be- fore S's were run their teachers were asked to predict their performance under the different reinforcement conditions. Subjects Sixty children from four Parochial Kindergartens served as subjects. These classes were from two Catholic schools in central Chicago. Kindergarten children were used be- cause most of them are old enough to understand verbal in- structions and young enough that inexpensive candy and trin- kets can be used effectively as rewards. The ages of the fijs ranged from 4 years, 9 months to 6 years, 2 months. There is some suggestion (Gerwitz & Baer, 1958a; Steven- son, 1961) that girls may reSpond more to approval comments from a male‘g than boys. To control for possible sex ef- fects an equal number of each sex was assigned to each experi- ymental group. The tangible, social, and alternating reinforcement groups were each divided into conditions of high (Fixed ratio 12 13 of l reinforcement per 10 reSponses, FRlO) and low (Fixed ratio of 1 reinforcement per 20 responses, FR20) density reinforcement. The rationale for using FRlO and FR20 sched- ules of reinforcement rests on the research of Long, Hammack, & Cambell (1958) on tangible reinforcement of children's bar pressing behavior. Using a variety of FR schedules they found that small ratios produced rapid satiation while ini- tial large ratios were often aversive. The FRZO was chosen as a ratio that might produce the least satiation while maximizing the probability that every subject would receive at least one reinforcement within each 20 sec. reSponse per- iod. The FRlO schedule of reinforcement was included to control for possible interaction effects between satiation and kind of reinforcement. Five girls and five boys were assigned to each of six eXperimental groups: Tangible Reinforcement Group T - high density (FRlO) N 10 (5 girls, 5 boys) Group T - low density (FRZO) N 10 (5 girls, 5 boys) Social Approval Comments Group S - high density (FRIO) N Group 8 - low density (FRZO) N 10 (5 girls, 5 boys; 10 (5 girls, 5 boys Alternation of Tangible Reinforcement with Approval Comments Group A - high density (FRlO) N = 10 (5 girls, 5 boys; Group A - low density (FRZO) N = 10 (5 girls, 5 boys The assignment of subjects may be summarized in terms of a 3 (kind of reinforcement) X 2 (density of reinforcement) X 2 (sex) design. 1 Fa.” ._‘ i 14 Apparatus The task was pressing a telegraph key. There were several reasons for selecting this task. In most cases 1 high, stable reaponse rates can be quickly obtained in reSponse to instructions. A very simple motor reSponse is required that is not physically incompatible with looking at the reinforcing stimuli. And, the task is relatively '. boring. There is little about the task itself that might h reinforce further reSponding. The apparatus (Figures 1, 2, 3, and 4) consisted of a 35" x 21" table with a 2" high step 6%" back from the front .- W‘r". ‘ ‘7‘ I. edge. A telegraph key (Johnson Standard 114-320) was mount- ed on a wooden paddle which g could manipulate to expose or withdraw the key through a 2" x 1%" hole in the center of the step. Screens (21" x 21") were mounted at each side of the table and two 17%" x 21" sliding screens were placed 12%" back from the front edge of the table. One of these screens could be moved to eXpose‘g. The other could be moved to expose a screen with a protruding tube through which candy (a mixture of M & M's, candycorn, "Good 'n Plentys", and jellybeans) could be drOpped into a clear plastic cup. The entire apparatus was white except for the black key, clear plastic cup, and a red line over the candy delivery tube. A stOpwatch was used to time the intervals of key ‘presentation. An electrical counter (General Controls CE4OBS402) kept a cumulative record of key pressing. The counter was insulated but still made a small audible click 15 Fig. 1. Apparatus During Rest Periods. Fig. 2. Apparatus During Base And Non-Reinforcement Periods. 16 Fig. 3. Apparatus During Tangible Reinforcement Periods. Fig. 4. Apparatus During Social Reinforcement Periods. 17 with each key press. ,g recorded the cumulative frequency after each interval of key presentation. Procedure The apparatus was set up in an unused room of each school. The children in each class were told thatlg had a list of their names and he would call each of them to come and do something for him. g, a male adult, walked S's to and from the classrooms and attempted to keep verbal inter- action to a minimum by reSponding as briefly as possible to questions from the children and otherwise not initiating conversation except as delineated by the procedure. Instructions. Each g was told to "Sit here!" Standing beside the apparatus,‘§ pointed to the chair in front of the key pressing apparatus and said, "See the black button? It goes in and out like this." ‘5 demonstrated by twice withdrawing and presenting the key. ,g said, "When it is out I want you to push it down and up like this!" g demon- strated by pushing the key 8 times at a rate of two per second. .E said, "Now you do it." E presented the key and after 5 presses withdrew the key while st hand was still on it. It was presented again and withdrawn after § made 8 more reSponses. If‘g failed to reSpond g said, "When the button is out, I want you to push it down and up." ,g made no approval comments in this condition. Base period. A base rate was established for each‘g in res- ponse to instructions only. After the instructions above l8 and with the key withdrawn E’said, "Remember, when the but- ton is out I want you to push it down and up." g sat down behind the screen and presented the key for four 20 sec. :_ intervals (Fig. l) separated by a "rest period" of 10 - 15 sec. during which the button was withdrawn (Fig. 2). E made no approval comments in this condition. The major reason for including the "rest period" be- tween trials was a wish for rapid discrimination between tangible and social reward conditions by the experimental group under alternating conditions of reinforcement. Where a high-rate response follows a low-rate reSponse the low-rate Irrfilfl'uuu a reSponse may increase in frequency (Premack, 1959). The "rest period" was included in order to minimize this effect. For comparable controls the "rest period" was held constant for all groups and all experimental manipulations. Reinforcement period. Following the period for establishing a base rate for the telegraph key, twenty 20 sec. presen- tations of the key were given under various conditions of F reinforcement. All key presentations were preceded and followed by a 10 - 15 sec. rest period, except where in- structions were given, as described below. The groups were rewarded in the following ways: 1) Group T (Tangible Reinforcement) - After the 10 - 15 sec. rest period following the base periodlg Opened .the panel revealing the reward chute and plastic candy con- tainer (Fig. 3). g said, "See the plastic cup. Candy can drop in the cup, like this." ‘E demonstrated. "When you see 19 the cup you can win candy by pushing the button down and up. After we are all finished today, just before I take you back to class, I will put the candy in a bag for you. If you win 1 enough candy you can trade it for some nice toys." The pan- A e1 was then closed. The panel was reOpened and the key pre- sented for 20 sec. Group T - high density, was rewarded on FR10 and Group T - low density, on FR20 schedules of rein- forcement. The button was withdrawn after the 20 sec. per- iod and the panel closed. After a 10 - 15 sec. rest period this procedure was repeated. The procedure remained the same through 20 trials. 2) Group 3 (Social Approval Comments) - After the 10 - 15 sec. rest period following the base period in response to instructions the panel was Opened revealing g and the key was presented for a 20 sec. interval (Fig. 4). Group 8 - high density, was rewarded by social approval comments on FR10 and Group S - low density, on FR20 schedules of rein- forcement. The comments were those used by Patterson & Hinsey (1964) and were said in the same sequence for each subject: "You're doing fine; M-hm; Good; Very nice; Very good; Fine; That's good; That's fine; Very good;" (Repeat sequence). After each 20 sec. interval the key was with- drawn and the panel closed. After a 10 - 15 sec. interval the procedure was repeated. This procedure remained the same for 20 trials. 3) Group A (Alternation of Tangible Reinforcement and Social Approval Comments) - After the base period this group received 10 trials under the procedure described for Group T 20 under which tangible reinforcement was presented for key pressing during the 20 sec. intervals of key presentation. These trials were alternated with 10 trials under the pro- 1. cedure described for Group S under which social approval comments were made for key pressing during the 20 sec. in- tervals of key presentation. Group A - high density, re- ceived an FR10 schedule through all alternations. For half of this group the alternations started with tangible rein- forcement. The other half started with social approval comments. Group A - low density, received an FR20 schedule of reinforcement. Alternations started with tangible rein- forcement for half the group and with social approval com- ments for the other half. Non-reinforcement period. After the 10 - 15 sec. rest per- iod following the reinforcement periodlg presented the key for eight 20 sec. intervals preceded and followed by 10 - 15 sec. rest periods. In this condition both sliding panels remained closed concealing g and the candy cup (Fig. 2). E g . did not Speak during this period. After the Non-reinforcement condition the candy was bagged or traded for small trinkets and each child was led back to his classroom. Teacher Ratings The button pushing task and reward conditions were des- ycribed to the teachers Of the four kindergarten classes be- fore the experiment and they were asked to predict the 21 performance of their children. They received a set of rat- ing sheets covered by the following instructions (Appendix ‘ A). "Estimate how well you think each child would perform :- when asked to do a boring task like pushing a button down I over and over again. Rate each child on cooperation, per- sistence, responsiveness to social approval and reSponsive- ness to tangible rewards." The rating sheets consisted of a list of the children's names. By each name was a scale from 1 to 7 on which the children were rated. There were separate sheets for "Coop- erates with adult instructions", "Persistence - performs well throughout an extended period of time", "Responsiveness to adult approval - Performance increases when an adult says encouraging things like 'You're doing a good job'; 'Very good'; etc."; and "ReSponsiveness to tangible rewards - Per- formance increases when encouraged by tangible rewards, such as candy and toys". For each scale the teacher was asked to give the most cooperative, persistent, or reSponsive child a 7, the least a 1, and then rate the remainder of the children in her class along the full 7 point scale. RESULTS Analysis of Reinforcement Effects A common way to analyze data from this kind of experi- ment is in terms of difference scores (Stevenson & Hill, 1966). Because there is usually a large variance between subjects in reSponse rates and since this kind of experiment is primarily concerned with variance due to reinforcement effects, the initial base rate for each subject is subtracted from his rates under the experimental conditions and the analyses are performed on the resulting difference scores. In this analysis the mean of trials 3 and 4 of the base period was used as an index of the base rate for each sub- ject. The correlation between Trial 3 (a = 47.6, N = 60) and Trial 4 (i = 46.7, N = 60) indicates a high degree Of individual consistency (r a .90, p<<:.0005). Except for the group under alternated reinforcement conditions, mean response rates were computed for each sub- ject over blocks of 4 trials in the experimental periods. This results in means for 5 blocks of trials under the Rein- forcement Period and 2 blocks of trials under the Non-Rein- forcement Period. Because the alternated group received 10 trials of social reinforcement and 10 trials of tangible reinforcement under the Reinforcement Period and these con- ditions were to be analyzed separately, means were computed ‘for each subject over blocks of 2 trials. For the alter- nated group in the Reinforcement Period there are, therefore, 5 block means computed from the tangible reinforcement 22 23 situation and 5 block means computed from the social rein- forcement situation. Because the 5 blocks during the rein- forcement period are based on 2 trials for the alternated conditions and 4 trials for the other conditions, a check was made on the reliability of the trial scores within blocks to see if a valid comparison could be made between blocks of 2 and blocks of 4 trials. The average inter-trial correlation within blocks is .90 for the alternated group under tangible reinforcement and .90 for the alternated group under social reinforcement. The average inter-trial correlation within blocks for the other groups during the Reinforcement Period was calculated two ways. The mean cor- relation between successive trial scores within blocks is .84. The mean correlation between the average of the first 2 scores and the average of the second 2 scores within blocks is .89. The average of inter-trial correlations within blocks of 2 trials for the alternated conditions are high enough that these blocks can be compared with the blocks of 4 trials under the other reinforcement conditions. Each subject's mean base rate was subtracted from the means of his 5 blocks of trials in the Reinforcement Period and the means of his 2 blocks of trials in the Non-Reinforce- ment Period. The analyses were then performed on these dif- ference scores. The mean difference scores for the experi- mental groups under the Reinforcement and Non-Reinforcement ‘Periods are plotted in Fig. 5. .ucmEOONONGAOmucOG paw unmEOOHOMGAOM mo mcowuwpcou ca ucoEmouowcamm mo mpcax unou0mwao Hops: manouu mo cowaumdeoo .m ouanm N a n a m N a 3:63 Sara b b b b b m 2- _ Amaoom poumchoua< u m< #3.. a. \ _ «SEEN Bumfimfia - ea ucoEOOH0mcHom poumcumua< u < udeoowowcaom Hmaoom u m unoEoou0wsamm OAOHmeH u H 9393 9393 mox; Burssala flax u; seouaxaggrq UQQEQUHOMGHMMICOZ _ USMEMUHOMCH 0m . .3 25 Comparison of Tanggple. Socia;i_and Alternated social rein- forcement. An analysis of variance was performed on the difference scores under the Reinforcement Period for the Tangible Reinforcement group (T), the Social Reinforcement group (S), and the Alternated group under the social rein- forcement condition (AS). The results appear in Table 1. There is a significant effect across trial blocks (F = 8.01, g; = 4, 192, p<<:.01). Inspection of Fig. 5 shows a gen- eral increase in response to reinforcement for all groups, A, T, and AS, with trial blocks. No significant differences were found between rein- forcement groups although it can be seen in Fig. 5 that the relative performance of T and S is in the«predicted direc- tion (T2=’S). There is no suppOrt for negative contrast effects or biphasic effects in the group comparison. How- ever, examination of individual performance indicates the possibility of such effects in some subjects who were ex- posed to both social and tangible reinforcement. In Fig. 6 the performance of 2 subjects is plotted to illustrate the possibility of contrast effects in some in- dividual cases. Subject #48, a boy, received reinforcement on an FR20 schedule with alternations beginning on social reinforcement. Subject #54, also a boy, received reinforce- ment on an FR10 schedule with alternations beginning on tan- gible reinforcement. Both subjects show an initial rise above baseline in reSponse to social approval comments fol- lowed by a sharp decline well below baseline. The increase Table 1 Analysis Of Variance Alternated Social Reinforcement ' Source of Variation df MS F .- ,1 Between subjects 22 . :V A (reinforcement) 2 448.10 1.69 3 B (density) 1 786.40 2.96 ' C (sex) 1 591.40 2.23 AB 2 103.30 .39 AC 2 232.80 .88 L BC 1 17.90 .07 ABC 2 604.10 2.27 Subj. w. groups 48 Withip subjects 240 0 (trial blocks) 4 258.93 8.01** AD 8 36.16 1.12 BD 4 27.10 .84 CD 4 183.43 5.68** ABD 8 17.39 .54 ACD 8 15.19 .47 BCD 4 10.43 .32 ABCD 8 31.036 .97 ' stubj. w. groups 192 32.32 **p <.01 27 ”I Reinforcement | Non-Reinforcement 3| l 60. I Tangible I I I 40« I I \I l I Social I 20‘ I I I I o I I , Periods341234567891‘0I2345678 Subject #48 Frequency of Key Pressing Within 20" Periods Social o I Periods34123456789101234.5678 Subject #54 Figure 6. Performance of Two Subjects Alternated Between Tangible and Social Reinforcement 28 in performance from the last trial under social reinforce; ment to the first trials under the Non-Reinforcement Period indicates that the approval comments may have served to de- press performance in later trials under social approval com- ments. As the group data indicates this pattern of perfor- mance was the exception and not characteristic of most subjects. There are no significant differences due to density of reinforcement and there are no significant interactions with this variable. Although there are no significant main ef- fects due to sex, the sex by trial interaction is signifi- cant (F = 5.68, pp = 4, 192, pg<:.01). Fig. 7 ilbustrates a greater increase in performance over trials for girls than for boys. An individual comparison between boys and girls across trial blocks 3, 4, and 5 is significant (F = 4.27, pg = 1, 48, p:<:.05) indicating that in later trials girls reSponded more to the conditions of reinforcement than boys. Comparison of Tangible, Social and Alternated tangible rein- forcement. An analysis of variance was performed on the difference scores under the Reinforcement Period for the Tangible Reinforcement group (T), the Social Reinforcement group (S), and the Alternated group under the tangible rein- forcement condition (AT). The results appear in Table 2. As in the above analysis there is a significant effect . across trial blocks (§.= 13.04, g§_= 4, 192, 2f<:.01) from the general increase in reaponse to reinforcement with trial blocks (Fig. l) for all groups. 29 +15‘ 0 «U a 3 +10J S E Girls 8 'H w +5- 5 m 0) O H n. 3‘ J x 0 5 Boys to O O 8 w ‘5' o N 'H 004 Q -10 1 r 17 f Trial 1 2 3 4 5 Blocks Figure 7. Performance of Boys and Girls During the Reinforcement Period for All Groups Table 2 Analysis of Variance Alternated Tangible Reinforcement Source of Variation Between subjects A (reinforcement) B (density) C (sex) AB AC BC ABC Subj. w. groups Wiphin subjects D (trial blocks) AD BD CD ABD ACD BCD ABCD stubj. w. groups *p<.05 **p'<:.01 MS F F 2.9. 2 1220.10 4.34* 1 1038.70 3.70 1 218.00 .78 2 46.30 .16 2 40.05 .14 1 38.20 .14 2 809.15 2.88 48 281.10 2‘19 4 376.85 13.04** 8 84.90 2.94** 4 11.43 .40 4 142.05 4.91** 8 17.55 .61 8 15.13 .52 4 53.70 1.86 8 52.29 1.81 192 28.91 31 In this comparison there is a significant effect be- tween reinforcement conditions (§,= 4.34, g; = 2, 48, 2'<:.05) and a significant interaction between trial blocks and reinforcement conditions (F = 2.94, pp = 8, 192, 2‘<:.01). It can be seen from Fig. 5 that reSponse to rein- forcement under the AT condition begins around the same level as the S and T groups but increases at a more rapid rate with trial blocks. Individual comparisons across trial ’ FE blocks 3, 4, and 5 indicate a significant difference be- tween the group under the AT condition and group S (E = 8.13, g; = 1, 48,‘p<::.01). The comparison between the group under the A condition and group T (§,= 2.52, g; = 1, 48, T p‘pp) did not reach the .05 level of significance. Compari- sons of the increases in performance between trial blocks 1 and 5 shows that under condition AT there is a signifi- cantly larger increase than for group T (E = 27.64, g; = 1, 192, p<.01) or group S (E = 19.25, p; = l, 192, p<.01). This gives some evidence for a positive contrast effect. As in the first analysis there are no significant ef- fects on interactions due to density of reinforcement. There is no significant main effect due to sex, but the sex by trials interaction is again significant (E = 4.91, g§,= 4, 192, p<.01). Comparison of groups under Non-Reinforcement. An analysis I of variance was performed on the difference scores under the Non-Reinforcement Period for the Tangible Reinforcement group (T), the Social Reinforcement group (S), and the 32 Alternated group (A). The results appear in Table 3. The difference between groups is significant (5,: 3.28, g§_= 2, 48,.p<=:.05). On examination of Fig. 5 it can be seen that the performance of group T relative to base rate is well be- low groups A.and S. Individual comparisons between groups T and A (F a 5.16, pg = 1, 48, p<.05) and groups T and S . (F = 4.67, pg = l, 48, pg<:.05) indicate that T is signifi- F cantly lower than the other 2 groups. There is a significant interaction of reinforcement den- sity with trial blocks (E = 4.87, g; = l, 192, ps=:.05). In Fig. 8 it can be seen that performance of the low density group drops with trials while the high density group rises. Individual comparisons between the groups for trial block 1 and trial block 2 did not reach the .05 level of signifi- cance. All other effects and interactions are non-significant. Comparison pf Alternated Conditions of Reipforcement. An analysis was performed to compare differences in performance under the alternated social condition of reinforcement (AS) and the alternated tangible condition of reinforcement (AT). This was done by calculating the difference for each subject between his last 4 trials under tangible reinforcement and the last 4 trials under social reinforcement. ReSponse to tangible reinforcement was higher than social reinforcement. . The mean difference per trial was 7.38 responses (N = 20) and was significant (p = 3.65, g; = 19, 2f<:.005). Although reSponse to tangible reinforcement was greater for most 33 Table 3 Analysis of Variance Non - Reinforcement Soupce of Vpriation df M; F Between subjects 22 A (reinforcement) 2 1774.30 3.28* B édensity) l .10 .00 C sex) 1 182.50 .34 AB 2 96.80 .18 AC 2 68.35 .13 BC 1 22.60 .04 ABC 2 327.80 .61 Subj. w. groups 48 541.09 Withgp subjepts 239 0 (trial blocks) 4 25.20 1.29 AD 8 37.15 1.90 BD 4 95.40 4.87* CD 4 .10 .01 ABD 8 52.55 2.69 ACD 8 4.05 .21 BCD 4 6.20 .32 ABCD 8 1.25 .06 stubj. w. groups 192 19.57 *p<=:.05 34 0 - E11 230 High I m DenSity s U) o H a. >.m -1 1 ‘ G) u L-»— Saw a: a sum (0 3 3 u 2 1 Low 5 8 7 Density Law 044 U4 “-4 H a -3 ' . Trial Blocks 1 2 Figure 8. Performance of High and Low Density Reinforcement Groups During the Non-Reinforcement Period 35 subjects, examination of each individual's performance revealed two subjects who responded consistently higher to the social approval comments. Analysis of Teacher Ratings In one kindergarten class (N = 19) a teacher's aide provided an opportunity to test the reliability of the scale ratings. Correlations between the teacher and her aide were 5 = .665 (p<<:.0005) for OOOperation, ; = .810 (p<=:.0005) for persistence,‘; = .509 (p<<:.01) for res- ponse to approval comments, and,p = .133 (nasg) for res- ponse to tangible rewards. This indicates reasonable agree- ment on the first 3 rating categories, and poor agreement on the rating for reSponse to tangible rewards. Intercorrelations of the teachers' ratings (the aide's ratings were not included in these analyses) on the 60 sub- jects are presented in Table 4. They show a significant relationship between ratings for COOperation, persistence, and response to social approval comments. There is a sig- nificant low'negative correlation between persistence and reSponse to tangible reward and a significant low positive correlation between response to social approval comments and reSponse to tangible reward. Correlations between the performance of subjects on base trials 3 and 4 with teacher ratings are low but signi- ficant for OOOperation (g = .293, pg<:.05), persistence (E = .258, 2f<:005) and response to tangible reinforcement (5 = .259, p;<:.05) and non significant for response to 36 Table 4 Intercorrelations of Teacher Ratings 8 '8 88' pa? 0 H “d a: an 'o m rro mro (D H' O O H-O .. . a a 2:8 8 g ram mt» 3' o (in pan :3 m 0 O m 0 a 8 a m H :3 o. n m COOperation '- .73** .65** -.12 Persistence ‘- —— ,48** -,24* Response to Approval Comments — -— — ,24* Response to Tangible Reward ‘—— —_ ._ ._ *2g<:.05 **p_ <.0005 37 mxooAO Adana powwow unosmouomnammusoz assess sass» powwow ucoamouowsaom Ho.HUvo «* no.uuva_k Hoo.- noo.+ ooa. «as. eoo. ooo.- ooo. osmsom .9 cu .ammm o3... moo... 32.- 33... a2... ooo.- ooo.- 38.88 mm 8 .m 3 doom .m ooo.- ooo.- m2... moo... NE... omo... also... mofiumamuom Md ooo. ooo. ona.- ooo.- om~.- ooa.- oo~.- coaumuoaooo moa.a moo.c s¢m¢.u kmm¢.n «sunn.n sknmn.u smam.n oumsmm .H on .dmom So; So. o3. ooo.- ooo. m3... 85.. .< .m wwcwmmMm WM 3o. moo... m3. o8. moo. 3o: ooo.- masseuse—mm m mma.-. naa.u who. nmo.u «00.: moo.n Hoa.n noaumwonoou M, ”I m s o N a mmmmmmmummmmmmmuuun. mowoom mosmuowman mam mwcaumm wetness noozuom msoaumamuuou n magma 38 social approval comments (5 = .109, p‘pp). Correlations between ratings and difference scores for both the group under tangible reinforcement (T) and the group under social reinforcement (S) are found in Table 5. The only significant relationship is between ratings for reSponse to tangible reward and performance of group T in the Reinforcement Period and this relationship is in a negative direction. Since the rating for response to tan- gible rewards is so unreliable the meaning of this "signi- ficant" relationship is ambiguous. DISCUSSION Difference in Response to Tangible and Social Reinforcement Although the means were in the predicted direction the group eXposed to the tangible reinforcement situation did not reSpond in a superior manner to the group rewarded by social approval comments. The reason the differences in the means were not significant was the large variation between subjects in response to both of these two reinforcement con- ditions. Perhaps a difference could have been demonstrated by making the social reinforcement weaker such as Costello (1967) did by concealing her‘g's behind a screen or by mak- ing the tangible reinforcement stronger. In an attempt to be "fair" this author used the strongest social and tangible reinforcement he could think of for this eXperimental com- parison. If this experiment was a fair comparison it cer-‘I tainly does not give much support for the general use of tangible reinforcement systems in place of social reinforce- ment, at least with kindergarten children. -- In the group exposed to both tangible reinforcement and social approval comments response was greater to the tangi- ble reinforcement situation, however, this difference is probably not due to the absolute value of these two kinds of reinforcement, but rather an interaction effect which will be discussed below. It is important to note that even though the reaponse rate increased more under tangible rein- forcement for most subjects in this group, this was not 39 1:.inx. I‘ll)!" I 4O true for all subjects. ThWte kind of rein- forcement to be used in individual treatment situations can _......L._..——-._ h...‘ to various kinds of reinforcers. __ _ __—..———...-...— '-.—. 7‘ Contrast Effects By far the most interesting result of this experiment was the reSponse to tangible reinforcement when this condi- F} tion was alternated with contingent social approval comments. The increase in response rate over trials under this condi- tion was more than either of the groups under tangible or run.“ .1 I 5 u social reinforcement alone. This difference is difficult to explain as a positive contrast effect, particularly since the group under tangible reinforcement did not differ sig- nificantly from the group under social reinforcement. A weak argument could be made that even though the difference between these two groups was not significant there still may be a difference in magnitude that produced a contrast ef- fect in the alternated group. Because this difference would be so small it is doubtful if it could produce such a con- trast effect, as the contrast would be even larger than the difference between magnitudes of reinforcement. Another possible explanation is that since tangible reinforcement was alternated with social reinforcement there was less satiation under the tangible reinforcement condi- tion. This is a weak argument because there was no differ- ence between the high and low density schedules of rein- forcement during the Reinforcement Period. If there was 41 satiation under tangible reinforcement the low rate schedule should have been more effective. Perhaps variety of rein- forcement was the important factor, but if this was so res- ponse under alternated social reinforcement should have been superior to the reSponse of the group under constant social reinforcement. Perhaps the strongest explanation is the one prOposed in the introduction, that tangible reinforcement may be more potent because it has intrinsic value and at the same time may represent social approval. Even though appro- val comments were not paired with tangible reinforcement, exposure to social reinforcement by‘g on alternate trials may have strengthened the association of social approval and tangible reinforcement creating an additive effect. This effect may be related to the results of two ex- periments employing token systems in classroom situations. Kuypers, Becker, & O'Leary (1968) found minimal effects in experimental manipulations of their token system as con- trasted to quite dramatic effects reported by O'Learylgg‘gl. (1967). On examination of their failure Kuypers 35 p1. (1968) discuss a major difference in the implementation of the two programs. The teachers in the successful program had extensive training in the application of behavioral principles emphasizing the use of contingent social rein- forcement while the teachers in the minimally effective program had no such training. This suggests that in appli- cation of tangible reinforcement systems the social compo- nent may be a crucial factor. 42 Although there was no evidence for negative contrast or biphasic effects in the group data this does not exclude the possibility of Jekyll - Hyde effects for those indi- viduals where there is high regard for tangible reinforce- ment and relatively low regard for social reinforcement. In subjects #48 and #54 it is possible that the alternating tangible reinforcement first seemed to increase reaponding under social reinforcement and then to depress reSponding under social reinforcement. The fact that both subjects' performance increased from later trials under social rein- forcement to initial trials with no reinforcement may mean that in the later alternations social reinforcement had be- come aversive and served to depress the response rate in this condition. Generalization to the Non-Reinforcement Situation while the groups with an experimental history of social reinforcement remained above baseline after the shift to the non-reinforcement situation the superior response of the group under alternated tangible reinforcement was not main- tained after the shift. After the shift to non-reinforce- ment the reSponse of the alternated group was no different than the group that had received no tangible reinforcement. While the group receiving tangible reinforcement was no different in reSponse to reinforcement than the group on social reinforcement, after the shift to the non-reinforce- ment situation performance dropped well below baseline, and significantly below the performance of the groups which had 43 social reinforcement in their experimental history. A num- ber of behaviors were observed during this period that sug- gested frustration in several children from the tangible re- inforcement group. These behaviors included very heavy key pounding, polite requests for more tangible rewards, shouts at p that p was not being rewarded, and various attempts to Open the screen covering the candy chute. These behaviors were not observed in the other groups. Why was there such a decrease in the performance of this group and not in the groups which had eXperienced social reinforcement? There is a big difference in the stimulus qualities of social and tangible reinforcement. In the natural environ- ment a good deal of human behavior is probably maintained by intermittent social reinforcement. Because of this the soc- ial reinforcement condition was not highly unlike reinforce- ment conditions these children experienced outside the ex- perimental situation. In fact, the situation was quite simi- lar to a parent or teacher giving a child a new task along with a high rate of contingent attention and then leaving the child to perform the task by himself with only occasion- al adult attention needed to maintain the new response. In contrast the tangible reinforcement in this experiment, like many tangible reinforcement systems employed with children, is very unlike most reinforcement systems in the natural en- vironment. The shift from an experimental history of only tangible reinforcement to the situation of non-reinforcement was quite abrupt with less generalization than the shifts 44 for those children who had eXperienced an experimental history that had included social reinforcement. A close look at the effect of reinforcement schedules under the non- reinforcement situation gives some support for this inter- pretation. A high density reinforcement schedule is less like non- reinforcement than a low density schedule. Performance of the low density group decreases gently with trials under the non-reinforcement situation as would be expected in extinc- tion. However, performance of the high density group rises from the point at which the shift was made from reinforce- ment to non-reinforcement. It is quite possible that the rise is due to a decrease in a frustration effect associated with the point of shift to the non-reinforcement situation. Although this rise was not large enough to be significant, the interaction between the rise for the high density group and the drop for the low density group was significant. Since the high density reinforcement situation was more dif- ferent than non-reinforcement, this group may have reaponded with more frustration effects on the shift to non-reinforce- ment. The Dr. Jekyll - Mr. Hyde phenomenon described in the introduction may be more related to the shift from tangible reinforcement (inside the classroom) to non-reinforcement (outside the classroom door) than the alternation between tangible reinforcement and relatively high rate social rein- forcement delivered in the highly structured eXperimental II; III III I III 45 situation. If this experimental analogy is valid, reSponse level under tangible reinforcement situations may not gen- eralize very well to non-reinforcement situations, indeed, it may even produce frustration effects and serve to depress reSponse levels. Needless to say, in the IJR project we did not wait for the results of this experiment to increase the rate of contingent social reinforcement in the classroom and ‘ gradually shift the tangible reinforcement to increasingly Ia intermittent schedules. For those children who were not entirely "weaned" from tangible reinforcement but who were ready to go back to public school classrooms we devised daily and then weekly report cards backed by long term re- wards from parents to ease the transition back to conse- quences of the natural environment. Sex Differences In this study the girls' rate of increase in response to reinforcement was greater than the boys. In light of the interaction effects between sex of‘g and sex.of S's (Gerwitz & Baer, 1958a; Stevenson, 1961) the assertion cannot be made that girls reSpond better than boys to reinforcement. It could be that boys would have performed better with a female ,g. Recent eXperiments and reviews of the literature in this area (Rosenbaum, 1969; Dusek, 1969) show inconsistent find- ings and the possibility that a number of variables are in- volved in sex differences in reSponse to reinforcement sit- uations. 46 The major reason sex was included as a variable in this study was to control for the possibility of confounding in- teractions between sex and response to the different rein- forcement conditions. There were no confounding interactions. Teacher Ratings The intercorrelations between ratings indicates that the teachers in this study perceived strong common elements in ratings of cooperation, persistence, and reSponse to soc- ial approval. The low negative correlation between ratings of persistence and ratings of reSponse to tangible rewards means there is a slight tendency for teachers to perceive persistent children as reSponding less to tangible rewards and/or children who reSpond well to tangible rewards to be less persistent. The implications in this perception may be that tangible rewards are most useful for those children who are not "naturally" persistent. Initial response to instructions from an adult during the base period was related to ratings for cooperation, persistence, and reSponse to tangible reinforcement but the predictive value of the ratings are quite low. It is an interesting inconsistency that the one rating category that was predictive of reSponse to the reinforcement situation was the rating on which there was almost no agreement be- tween the teacher and her aide. The indication is that this teacher agreed more with the other three teachers on the meaning of response to tangible reinforcement than with her aide. What is surprising is that the prediction of reSponse 47 to tangible rewards has a negative relationship to actual reSponse of children to tangible reinforcement. In general, there was better agreement on teacher perception of the meanings of the ratings for their children than between teacher ratings and the actual performance of the children in the experimental situation. Teacher ratings were poor predictors of their children's performance. Maigr Conclusions The primary findings of this study are that for kinder- é garten children tangible rewards do not generally have more reinforcement value than social approval comments from an s~i adult. A combination of tangible and social reinforcement may be more effective than tangible or social reinforcement alone. ReSponse level under tangible reinforcement may not generalize to a non-reinforcement situation and under some conditions may cause more of a decrease in reSponse level than reSponses previously reinforced by social reinforce- ment. Although there are wide variations between individual children in reSponse to social and tangible reinforcement, teacher ratings are of little value in predicting the res- ponse of individual children. It appears that the application of Operant conditioning principles to treatment of childrens' behavior is not so simple as rewarding apprOpriate reSponses. There is wide variation in the way children respond to different kinds Of’ rewards and much thought must be given to the manner in which appropriate reSponses are shaped and strengthened so 48 that they can eventually be maintained by natural conse- quences. An important guideline of the IJR project has become a rule that no artificial structure is imposed as part of the treatment program without plans for shifting control of the behavior to the natural environment. B IBLIOGRAPHY BIBLIOGRAPHY Allen, K. Eileen, Hart, Betty, Buell, Joan 5., Harris, Florence R., & Wolf, M. M. Effects of social rein- forcement on isolate behavior of a nursery school child. Child DevelOpment, 1964, 2;, 511-519. Amsel, A. The role of frustrative non-reward in non- continuous reward situations. Psychological Bulle- tin, 1958, 22, 102-119. Amsel, A. Frustrative non-reward in partial reinforcement and discrimination learning: Some recent history and E I 8"“ a theoretical extension. Psychological Review, 1962, 92, 306-328. Armus, H. L. Effect of magnitude of reinforcement on acquisition and extinction of a runnin res onse. Journal pf Expergmental Psychology, 19 9, 2_, 61-63. Ayllon T. and Azrin N. The TOken Economy. Appleton - century - Crofts, 1968—. “'— Becker, w., Madsen, C. Jr., Arnold, G. & Thomas, D. The contingent use of teacher attention and praise in reducing classroom behavior problems. Journal pf Special Education, 1967, L, 287-307. Birnbrauer, J., WOlf, M., Kidder, J. & Tague, C. Classroom behavior of retarded pupils with token reinforcement. Journal pf Experimental Child Psychology, 1965, g, Zlg-ng. Bower, G. A contrast effect in differential conditioning. Journal pf Egperimental Psychology, 1961, pg, 196-199. Brown, P. & Elliott, R. Control of aggression in a nursery school class. Journal pf Experimental Child Psychology, 1965, g, 103-107. Brown, R., Pace, 2. & Becker, w. Treatment of extreme nega- tivism and "autistic" behavior in a six year Old boy. Exceptional Children, In Press. 49 50 Costello, J. Social reinforcers as functions of social competence and reinforcement conditions. Unpublished Pgég. dissertation, Illinois Institute of Technology, 1 . Dunham, R. Contrasted conditions of reinforcement: A sggegfgve critique. Psychological Bullet' , 1968, pg, 2 "’ o Dusek J. An investigation of experimenter bias in the per- formance of children at a simple motor task under con- ditions of social reinforcement and non-reinforcement. Egggblished Ph.D. dissertation, University of Illinois, Ferster, G. & DeMyer, M. The development of performance in autistic children in an automatically controlled en- yirogzgnt. gournal pf Chronic Diseases, 1961, 1;, 2" o Gerwitz, J. & Baer, D. The effect of brief social depriva- tion on behaviors for a social reinforcer. Journal 22 Abnoppal ppg Social Psychology, 1958a, pp, 49-55. Gerwitz, J. & Baer, D. Deprivation and satiation of social reinforcers as drive conditioners. Journal 2; Abnprmal ppg Social Psychology, 19586,‘21, l65-I72. Green, C. & Zigler, E. Social deprivation and the perfor- mance of retarded and normal children on a satiation type task. thld DevelOpment, 1962, 33, 499-508. Glass, 0. & Ison, J. The effects of incentive shifts fol- lowing differential conditioning. Paper presented at the meeting of the Psychonomic Society, St. Louis, Oct- ober, 1966. Harris, Florence R., Johnston, Margaret, Kelley, 0. Susan, & WOlf, M. M. Effects of positive social reinforcement on regressed crawling of a nursery school child. Jour- pp; __ Educagional Ppychplpgy, 1964, 5 , 35-41. Hart, Betty, Allen, K. Eileen, Buell, Joan S. Harris, Flo- rence R. & WOlf, M. M. Effects of sociaI reinforcement on operant crying. ipurnal p§.Experimentpl Child Ps - ChOlO , 1964. J‘, 14 " 0 Hill, K. & Stevenson, H. Effectiveness of social reinforce- ment following social and sensory deprivation. Journal 2; Abnormal and Social Psychplogy, 1964, pp, 579-584. 51 Hulse, S. H. Jr. Amount and percentage of reinforcement and duration of goal confinement in conditioning and extinc- Egoni Journal pngxperimental Psychology, 1958, pp, -5, Kohn, M. & Silverman, H. A competence scale and symptom checklist for the preschool child. Unpublished manu- fggépt. William Alanson White Institute, New York, Johnson, S. & Brown, R. Producing behavior change in parents of disturbed children. Journal pf Child Psycholpgy and Ppychiatpy, In Press. m3 Kuypers, D., Becker, W. & O'Leary, K. How to make a token system fail. Exceptional Children, 1968, 35, 101-109. Long, E., Hammack, F. & Cambell, B. Intermittent reinforce- ment of Operant behavior in children. Journal of the Experimental Analysis pf Behavior, 1958, L, 5:339. Lovaas, 0. 1., Berherich, J. P., Perloff, B. F., & Schaeffer, B. Acquisition of imitative Speech by schiZOphrenic children. Science, 1966, 151, 705-70 . Madsen, C., Becker, W. & Thomas, D. Rules, praise, and ignoring: elements of elementary classroom control. Journal pf Applied Behavior Analysis, 1968,‘;, 139-150. Marlowe, D. Need for social approval and the operant condi- tioning of meaningful verbal behavior. Journal pf Con- sulting Ps choloO , 1962, 26, 79-83. O'Leary, K. D. & Becker, W. C. Behavior modification of an adjustment class: A token reinforcement program. E5- ceptgonal Children, 1967, 3;, 637-642. Patterson, G. & Hinsey, W. Investigations of some assump- tions and characteristics of a procedure for instru- mental conditioning in children. Journal pf E§peri- mental Child Ps cholo , 1964, L, III:I22. Perce, Frances. Use of behavior modification techniques with six boys and their parents. Paper presented to the Frances J. Gerty Awards Program, Illinois Depart- ment of Mental Health, June, 1968. Pieper, W. & Marx, M. Effects of within session incentive contrast on instrumental acquisition and performance. Journal‘pf Engrimental Psychology, 1963, pp, 568-571. 52 Premack, D. TOward empirical behavior laws: 1. Positive reinforcement. Psychological Review, 1959, pp, 219-233. Rosenbaum, M. A methodolo ical investigation of social rein- forcement studies: t e effects of fixed-interval and fixed-ratio schedules on time and rate measures. Un- pgggished Ph.D. dissertation, University of Illinois, Spielberger, C., Berger, A. & Howard, K. Conditioning of verbal behavior as a function of awareness, need for social approval, and motivation to receive reinforce- ment. ournal of Abnormal d Soci 1 Ps cholo 1963 22., 241%...___._aa.__a._z__sx. . Stevenson, H. Social reinforcement with children as a func- tion of CA, sex of E. and sex of S. Journal pf Abnor- mal and Social Psychoiogy, 1961, pg, 147-I54. Stevenson, H. & Fahel, Leila. The effect of social rein- forcement on the performance of institutionalized and non-institutionalized normal and feebleminded children. Journal pf Personality, 1965, 33, 418-427. Stevenson, H. & Hill, K. The use of rate as a measure of reSponse in studies of social reinforcement. Psycho- logical Bulletin, 1966, pp, 321-326. Stevenson, H. & Odom, R. The effectiveness of social rein- forcement following two conditions of social depriva- g§ona3140urnal pf'Abnormal Egg §pc§al Psycpplogy, 1961, —’ ° Thomas, 0., Becker, W. & Armstrong, M. Production and eli- mination of disruptive classroom behavior by systema- tically varying teacher's behavior. Journal p£.Applied Behavipr Anal sis, 1968, 1, 35-45. Wagner, A. R. Effects of amount and percentage of reinforce- ment and number of acquisition trials on conditioning and extinction. gopzpal 2; Egperimental Ps cholo , 1961, 6 , 234-242. wahler, R. Winkel, G., Peterson, R. & Morrison, D. Mothers as behavior therapists for their own children. Behavior Research and Therapy, 1964, 3, 113-124. Wolf, M., Giles, D. & Hall, V. EXperiments with token rein- forcement in a remedial classroom. Behavgor Research and The a , 1968,‘§, 51-64. 53 WOlf, M., Risely, T. & Mees. Application of Operant condi- tioning procedures to the behavior problems of an au- gigtgc child. Behavior Research and Thera , 1964, L, 0.12. . . Hr} IT; I III . .0 . . ; Ipstltklli \ . PIKE'waflraliu. . .. . . . .v 4b.»... |--|' “I“. *‘fl‘ f... b I --.r - “ms-W1. APPENDIX ' ’v‘.‘ .‘. V “.2, V 1 54 Estimate how well you think each child would perform é"" when asked to do a boring task like pushing a button down over and over again. Rate each child on cooperation, persistence, responsiveness to social approval and reSponsiveness to tangible rewards. 55 COOperates With Adult Instructions Give the most cooperative child a 7. Give the least cooperative child a 1. Then rate the rest of the children. Least 11935 W W 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 56 Persistence - Performs Well Throughout an Extended Period of Time Give the most persistent child a 7. Give the least persistent child a 1. Then rate the rest of the children. PeFEIEFent PengEFent 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 57 ReSponvaeness to Adult Approval - Performance Increases When an Adult Says Encouraging Things Like "You're Doing a Good obz". "Vepy Goodj", etc, Give the child most reSponsive to adult approval a 7. Give the child least reSponsive to adult approval a 1. Then rate the rest of the children. 8228.2. 8.28.; ReSponsive Reaponsive l 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 58 Responsiveness to Tangible Rewards - Performance Increases When Encouraged by Tangible Rewards, Such as Candy And Toys. Give the child most responsive to tangible rewards a 7. Give the child least responsive to tangible rewards a 1. Then rate the rest of the children. seas; 892.2 ReSponsive Responsive 1 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 “IIIIIIIIIIIIII IES II