The effect of vicarious punishment on prosocial behavior in children

The effect of vicarious punishment on prosocial behavior in children

JOT;RNAI, OF EXPERIMENTAL The Effect WILLIAM CHILD of Vicarious Behavior If. 15, 222-236 PSYCHOI,O(:T MORRIS,~ AND ROBERT Dartmouth Punish...

983KB Sizes 0 Downloads 67 Views

JOT;RNAI,

OF EXPERIMENTAL

The

Effect

WILLIAM

CHILD

of

Vicarious Behavior If.

15, 222-236

PSYCHOI,O(:T

MORRIS,~ AND

ROBERT Dartmouth

Punishment in Children1 HALINA

S.

11.

(1973)

on Prosocial

MARSHALL,

MILLER

College

Two experiments are reported dealing with the effects of vicarious punisbment on prosocial behavior. In Experiment I, first and second grade females who witnessed a peer model being punished for a refusal to share (“nonsharing, punished”), shared more than an appropriate control group. Sharing was similarly heightened, however, in a group who witnessed punishment to the model which was not contingent upon any specific behavior (“punishment only”). Experiment II t&cd and confirmed the hypothesis derived from Experiment I that noncontingent vicarious punishment has a generalized inhibitory effect on antisocial behaviors. First, second, and third grade females were shown either the “nonsharing, punished,” “punishment only,” or baseline videotapes used in Experiment I. Subsequent to viewing the condition helped the experimenter tapes, subjects in the “punishment only” significantly more than did subjects in the “nonsharing, punished” and baseline conditions. The ability of existing theoretical treatments of vicarious punishment to account for these results is questioned.

Several researchers have studied the effects of exposure to altruistic models and positive vicarious reinforcement on sharing behavior (e.g., ElIiott & Vasta, 1970; Harris, 1970; Midlarsky & Bryan, 1967). Significant increases in the incidence and amount of sharing by observers have repeatedly been found as a function of such treatments. On the other ‘Experiment I is based on a portion of a dissertation written by the second author under the supervision of the first author and submitted to Dartmouth College in 1972 in partial fulfillment of the requirements for the Ph.D. degree. The authors gratefully acknowledge the helpful advice of Rogers Elliott, Robert Fleck, and John Lanzetta. In addition, the authors wish to express their appreciation to Mr. Millard Harrison, Superintendent of Schools in White River Junction, Vermont ; Mrs. Alice Smith, Mrs. Shirley Jardin?, Mr. Douglas Carver, Mr. Carl Van de Bogart, and Mr. William Brigham, the principals of schools in Hartford, Vermont, White River Junction, Vermont, Lebanon, New Hampshire, West Lebanon, New Hampshire, and Enfield, New Hampshire for their cooperation in securing subjects for these experiments. * Requests for reprints should be addressed to William N. Morris, Department of Psychology, Dartmouth College, Hanover, NH 03755. Copyright All rights

222 @ 1973 by Academic Press, Inc. of reproduction in any form reserved.

VICARIOUS

223

PUNISHMENT

hand, the effectiveness of vicarious punishment of a model’s antisocial behavior has not been adequately explored. If Bandura (1969, pp. 30-31) is correct in suggesting that vicarious reinforcement provides information about the behavioral dernands of the immediate social environment, then punishment of an antisocial response of a model should provide information roughly equivalent to that offered by rewarding a model’s prosocial response. In line with this reasoning, the central purpose of Experiment I was to test the general hypothesis that vicarious punishment of t.he failure to share would increase sharing behavior. To this end, children’s sharing behavior was examined as a function of observing a filmed model in one of four conditions3 in which: (1) the model refuses to share and is verbally punished for it (nonsharing, punished) ; (2) the filmed interaction involves no sharing or punishment (baseline) ; (3) the film ends immediately after the model refuses to share (nonsharing, no outcome) ; (4) the model is punished only, there being no sharing behavior involved (punishment only). The major prediction is that sharing behavior will be greater in children who observe a nonsharing model being punished as compared to children in an appropriate control group. The consequences to the model should serve as a cue signifying the nonpermissibility of the punished response .within a given social context. If tL hypothesis is empirically supported, the theoretical interpretation remains ambiguous. This is the case because observing a nonsharing act being punished inyolrcs two separate components each of which might be sufficient, from a theoretical point of view, to account for an increase in sharing. As for the possible effect of nonsharing alone, while a number of studies (e.g., Elliott & Vasta, 1970; Harris, 1970; Walters, Parke, & Cane, 1965) have shown that the mere occurrence of a prosocia’ behavior on the part of a model is sufficient to induce that behavior, it is also possible that the observation of an antisocial behavior enacted by a model can also induce prosocial behavior in observers. Such a prediction might be made by Berkowitz’ social responsibility norm or by Bandura’s concept of self-administered standards. According to Berkowitz (Berkowitz, 1968; Berkowitz & Daniels, 1964)) increasing the salience of the social responsibility norm tends to bring behavior more in line with its dictates.’ One way to make the norm more salient is to draw at’tention to the dependency of a potential recipient of help. In this context, the observation by a child of a model’s refusal to

The

3 There are other data not directly interested reader mav consult

related Marshall

to the central (19’72’2).

concern

of Experiment

I.

224

MORRIS,

MARSHALL,

AND

MILLER

share her candy with “a little girl who does not have any” could make more salient to the observer the dependency of those who have no candy on those who possess such resources. If this is true, then the nonsharing act itself, independent of punishment, could elicit. increased sharing behavior. According to Bandura (1969), on the other hand, while social behavior is controlled at the beginning mainly by external consequences of the child’s behavior, it eventually comes to be regulated by covert self-reinforcing operations which rely on symbolically generated consequences in the form of esteem-enhancing reactions. A person presumably compares his own or others’ behavior with his internalized standards and judges it positively or negatively depending on whether the performance exceeds or falls below the standard. By this interpretation, selfish behavior of a model could be evaluated negatively and, thereby, inhibited in the observer. In order to evaluate the possibility that the observation of nonsharing, per se, increases sharing behavior, one condition of the experiment included a model who refuses to share following which the film is immediately terminated with no consequences to the model. The second component of the complex stimulus situation is l)unishment and there is reason to believe that it, alone, might also increase sharing behavior. Given that a child witnesses a model punished, this might signal to the child that the situation is potentially dangerous. Such information might induce anxiety which, in turn, might lcad the child observer to avoid behaviors which, in his personal history, have led to punishment and to substitute socially desirable behaviors. In the particular situation in the present study, a child, when given the choice brtwecn sharing or not, may share in order to “put his (her) best foot forward” in a demonstrably “dangerous” environment. In order to test this notion, one condition of the experiment included observation of a model who is punished without exposing observers to the antecedent behavior which presumably elicited the punishment, FinalIy, Bandura’s treatment of observational Iearning depends to a great extent on the notion that outcomes to a model provide the observer with information about what is or is not appropriate in a particular social oont‘ext. It follows from this view that the effects of observed behavioroutcome sequences should vary as a function of the similarity of such sequences to the observer’s situation. To the extent, then, t.hat the social context for the model and observer are highly similar, the observed behavior-out’come contingency is a more valid predictor of what the observer might experience. In order to subject this notion to an experimental test, two parallel sets of modeling films were used with two different socializing agents.

VICARIOUS

PUNISHMENT

225

Half of the children in each modeling condition were exposed to an experimenter who was identical to the socializing agent they saw on the film, while the other half were exposed to a socializing agent who was different from their experimenter. It was predicted that directional differences from baseline, i.e., more or less sharing, in t.he various modeling conditions would be greater when the socializing agent and experimenter were identical than when they were different. METHOD

Subjects One hundred and twelve first and second grade girls, drawn from three local elementary schools, participated in the experiment. All three schools are located in small towns which are predominantly lower middle class. Average age of the group under study was 7 years, 5 months. There mere eight experimental groups (4 modeling conditions X 2 socializing agents) with 14 subjects in each condition. Equipment

and Apparatus

The experiment was conducted in the children’s own schools, in small rooms equipped wit,h a desk, small table, and several chairs. The apparatus used in the study consisted of: (a) TV camera, Shibaden model No. HV-14, soligor television lens 25 mm, f 1.4; (b ) Sony i/z in. TV monitor CVM-180UA, 18 in.; and (c) Sony 1/e in. videorecorder, model cv-2000. A TV monitor and vidcorecorder were used to present the modeling situations. Each subject was allowed to see herself on the television monitor at the beginning of the session. The purpose of this maneuver was to attract the subjects’ attention to the TV equipment and to help them to become accustomed to the experimental situation. An experimental assistant was employed to operate the video equipment. He remained in the experimental room at all times and did not interact in any way with the subjects. Modeling

Conditions

The design of the experiment was 4 X 2, with four modeling conditions and two socializing agents portrayed on tape. The first part of the videotape was identical in all conditions: it shows a 7-year-old girl sitting on the floor and playing with t,oys. An adult female (socializing agent) enters the room, asks the model for her name and how she likes playing with the toys. In the baseline condition, following this introductory interaction, the

226

MORRIS,

MARSHALL,

AND

MILLER

model continues to play for the rest of the film while t.he adult sits at the table and starts reading a book. In all other conditions, the adult offers the model a big bag of candy. punished” conIn the “nonsharing, no outcome” and the “nonsharing, ditions, the model is told that she may keep the candy for herself or share it with another (fictitious) girl who does not have any candy. The model announces that. she will keep all the candy for herself. In the “nonsharing, no outcome” condition the film terminate:: immediately after the model’s refusal to shurc. In the “nonsharing, punished,” condition the adult disapproves of the model’s behavior, announces t’hat she will not let the model play with the toys anymore, and asks her to leave the room. In the “punishment only” condition, subsequent to receiving the bag of candy the video-tape is faded out. After a short while, the video-tape fades in to show the socializing agent sitting alone at a table. She gets up, goes off camera, and returns with t.he model. At this point the socializing agent administers the identical punishment as in the ‘Lnon-sharing, punished” condition. There wcrc two parallel sets of video-tapes made with two female adults as socializing agents. The experimenter served as a socializing agent in one of these tape sets, while the other socializing agent was played by a psychology graduate stuclcnt of the same age as the experimenter. PROCEDURE

Subjects were brought individually to the experimental room. They were assigned to one of the eight experimental conditions by order of their arrival at the experimental room, with the sequence of treatments being randomly determined. The suhjcct and experimenter sat at the small table in front of a TV camera and monitor. The subject was asked her name, grade, and the name of her teacher. Then the subject’s attention was called to the television monitor with the subject’s image on it, and after a few seconds, she was asked to watch anot,her girl on TV. At this point the experimenter announced that she had to leave for a while, but that she would be back soon and left the room. In this way the experimenter remained ignorant of the subjects’ assignment to the treatment’ conditions throughout the experiment. The experimental assistant presented one of the video-taped modeling conditions and called the experimenter back to the room at the ronclusion of the video-tape. The experimenter thanked the subject for roming, gave her a bag containing 100 M&MB, and asked the subject to write her name on the bag. The experimenter then said that there would not he enough candy for all the children who were to come and suggestedthat perhaps the subject would

VICARIOILS

227

PI-NI8HMENT

want to leave some of her candy for children coming later. The experimenter pointed to a bowl on the table and said that, the subject could either put some in the bowl and that this candy would be given to the other children, or she could keep all of her cuudy for herself. At this point the experimenter stopped paying attentiou to the subject and busied herself with the lists of subjects’ names which were on the table. If the subject hesitated the experimenter repeated the part of the instructions emphasizing that it was ~11) to the subject to decide what to do with her candy. In each school, after several subjects had been run, randomly selected children were asked if they knew what other children were doing in the experimental room. None of the children sampled reported knowing any more than that viewing of tclerision was involved. While the experimenter was taking one subject to the classroom and bringing back another one, the experimental assistant prepared the videotape equipment for the presentation of the next modeling tape, and counted and put away the candy the subject had left in the bowl. Each child spent around 5-7 min in the experimental room, of which time approximately 2.5 min were spent watching t.he modeling tape. RESULTS

Sharing behavior as indicated by the number of M&MS children gave away was the dependent measure. Examination of the data showed that the sharing behavior demonstrated in this experiment was not normally distributed, and that the departure from normality was extreme. Since the form of the distribution of the data, and the scale of measurement used do not meet the requirements for parametric analyses, the data were ranked and nonparamet,ric tests were utilized wherever possible in the analyses. The median number of candies shared as a function of modeling condition and socializing agent is reported in Table 1.

Experimenter Not experimenter

:i 0

a B = Baseline; N-9, N-S, P = Nonsharing,

NO = Nonsharing, punished.

7.5 12

no outcome;

17.5 27

31.5 28

P 0 = Punishment,

only;

228 Modeling

MORRIS,

MARSHAT>T,,

AND

MII,I,ER

Conditions

A Kruskal-Wallis analysis of variance by ranks revealed that there were differences among the four modeling conditions in the number of M&MS shared (H = 8.94, p < .05) when the data were collapsed across socializing agents, grades, and schools. Compared with the baseline, children in the “nonsharing, punished” and the “punishment only” conditions shared significantly more (Mann-Whitney I; Test, x = 2.567, and 1.991, p < .05, two-tailed). The only other significant difference was between the “nonsharing, punished” and the “nonsharing, no outcome” conditions (Z = 2.110, p < .05, two-tailed). Socializing

Agents

It will be recalled that the principal prediction made with respect to the use of different socializing agents was that. of an interaction between agents and the modeling conditions. Since nonparametric techniques for measuring interaction terms have not yet been developed, the data were transformed using the square-root, transformation and these scores were submitted to a two-way analysis of variance (2 X 4), with socializing agents X modeling conditions. The F ratio used to test significance has, among its assumptions, conditions which are not fulfilled in the present study. Consequently, the results of the analysis of variance have to be treated as approximate results and interpreted cautiously.4 The effect of modeling condition was significant (F (3,104) = 2.70, p < .05). However, neither the main effect of socializing agent nor its interaction with modeling conditions approached significance (F(1,104) = 0.57 and F(3,104) = 0.30, respectively). It should be pointed out, however, that an informal check on this manipulation instituted part of the way through the experiment revealed that not all subjects recognized the intended identity of, or difference between, the experimenter and socializing agent. It is possible that had wider differences in similarity been created, perhaps by changing the age and/or sex of the socializing agent, an effect might have been detected. On the other hand, it may be the case that. the effect of vicarious punishment overrides the operation of more subtle variables. Grades A Mann-Whitney U Test comparing first and second graders revealed significant differences in sharing behavior between these two grades ‘On the other hand, quist, 1956, p. 78) that experimental situations sumptions of normality

it has been demonstrated (Boneau, 1960; Norton, in Lindthe F test of the analysis of variance can be used even in in which there is serious doubt about the underlying asand homogeneity of variance.

VICARIOUS

229

PUNISHMENT

(z = 1.92, p < .06, two-tailed). Sharing in the second grade was systematically higher than in the first grade in all modeling conditions, except for the baseline condition where medians were zero for both grades. The hypothesis of a correlation of generosity with age has been supported by numerous studies (Bryan & Test, 1967; Elliott & Vasta, 1970; Handlon & Gross, 1959; Midlarsky & Bryan, 1967; among others). While not unexpected, the present finding concerning age differences is somewhat surprising in that such a correlation is usually found when comparisons are made between subjects from grades separated by several years. In fact, several investigators (Bryan & Walbek, 1970; Harris, 1970; Rosenhan & White, 1967) have failed to find differences in t,he level of generosity between children from consecutive grades. DISCUSSION

The findings of the present study support the hypothesis that sharing behavior, as indexed by amount shared, is higher in children who observe a selfish model being punished than in children in the baseline group. Results from the “nonsharing, no outcome” condition revealed that there was no significant effect on sharing as a consequence of merely observing selfish behavior. This tends to rule out of consideration any explanat,ion based solely on the premise that observation of behaviors which fall below a subject’s own internalized standards causes those standards to be somehow “activated” or “elicited,” with the result of increasing socially responsible behaviors. indicate a Results from the “punishment only” condition, however, significant increase in sharing behavior when compared with the baseline condition. Furthermore, the “punishment only” condition does not differ significantly from the “nonsharing, punished” condition. Within the constraints of t’he design, then, this pattern of results suggests that the observation of “noncontingent”” punishment to a model, in a context in which sharing or not sharing becomes an issue, is a sufficient condition for inducing increased sharing. There is a growing literature on the effects of direct noncontingent punishment, in animals. In general, that literature (e.g., Boe & Church, 1968; Campbell k Church, 1969) indicates that there is less suppression of a given response with noncontingent shock than with contingent shock, and that the suppression resulting from noncontingent shock is more transient. Although the magnitude of suppression is greater with con5The term “noncontingent” is meant to imply nothing mental operations. In that sense. the punishment that the not contingent upon any observable behavior of the model.

mow child

than model

the experireceives is

230

MORRIS,

MARSHALL,

AND

MILLER

tingent than noncontingent shock, the degree of general emotional upset of an animal seems to be greater with noncontingent shock. While it is tempting to extrapolate from this literature to the current experiment, there are a number of important differences which make generalizations hazardous. In the first place, such generalizations would have to he made across a broad phyIogcnet.ic gap, essentially from pigeon and rat, to man. While such a leap may not be inherently undesirable, there are indications that cognitive factors may play a significant role in human response inhibition, e.g., Walters and Andres (1967) and Parke (1969). A similar assertion would seem to be gratuitous in terms of explaining response inhibition in infrahumans. Secondly, many of the studies done with animals employ high intensities of electric shock. This iu to be contrasted with the present study which used a comparatively mild verbal punishment’. Thirdly, the studies done with animals typically extend over a period of time including a number of trials, while t,he present study involved the presentation of only one punishment. Finally, t,he punishment in the present, case was vicarious as opposed to the dirert punishment used in the animal studies. Taken together, all these diffcrrnces suggest) that emotion-based explanations which predominate in the literature on noncontingent shock applied to animals, may not bt entirely satisfactory in dealing with the present effects. A more fruitful line of argument dealing with the effects of vicarious noncontingent punishment might better begin with an examinat,ion of what the child learns during exposure to modeling sequences. In the bulk of experiments which have hcen carried out using vicarious reward or punishment the child csscntinlly discovers that a given behavior brings tither good or bad outromes to a model. On the basis of t’he findings from these studies, it would appear that children tend to assume that the “rules of the sit.uation” arc likely to al)ply to them also since they tend to reproduce those behaviors which arc rewarded and avoid those behaviors which are punished. Consider, in t,his context, the plight of a child in the L’punishment only” condition. She observes a model, similar to herself in age and sex, receive a strong verbal reprimand, an outcome t.he subject. no doubt would prefer to avoid. Unfortunately, though. there is no information available to the child as to exactly what hehavior triggered this punishment to the model, and, thcrcfore, no indication of what it is that she should do or not do. One possible solution for thcb child is to In-oducc only those behaviors which, in the past, have Ilad a hi& probability of hringin g adult approval or, alternat,ively, a low 1)robabilitv of incurring adult disapproval. One behavior which surely possesses these properties is sharing. Thu P, when put in a situation where there is a choice between sharing and not. sharing, sharing is increased.

VICARIOUS

231

PUNISHMENT

A testable implication of this analysis is that noncontingent punishment, as opposed to contingent, has more general effects in t.erms of making more likely a whole range of behaviors, namely those behaviors that the child regards as socially appropriate or, more simply, the behaviors for which the child has received social approval in the past. Experiment II directly addresses t’his hypothesis. EXPERIMENT

II

While Experiment I demonstrated that the effect of showing children a film of a model being punished for selfish behavior was to increase sharing, the results of the “punishment only” condition make interpretation difficult. Since the “punishments only” condition produced essentially as much sharing as the “nonsharing, punished” condition, the argument could be made that in hot.1~ cases merely wit.nessing a model being punished produced the increase in sharing. The “punishment only” effect also calls into question the interpretation of those previous studies which demonstrated inhibition of specific behaviors following vicarious punishment of those behaviors. Bandura (1965)) for example, demonstrated that seeing a model punished for aggressive behavior inhibits imitative agressivc behavior of nursery school children. Walters, Lest., and Mezei (1963) and Walters, Parke, and Cane (1965) showed that viewing a model punished for breaking a taboo increased resistance to temptation in kindergarten children. None of these studies included a control group where children were exposed only to the punishment part of the film. Perhaps in all of these cases then, the effect of vicarious punishment was not to inhibit specific antisocial behaviors, but, as in the “punishment only” condition in Experiment I, to arouse a very generalized caution in the subjects, put them on their guard, and make antisocial behaviors less likely. It is perhaps more reasonable, however, to argue t’hat while seeing noncontingent punishment creates such a general response set, seeing specific antisocial acts punished only inhibits those specific acts. Experiment II tests the hypothesis that viewing the “punishment only” tape of Experiment. I will increase performance of a prosocial behavior other than sharing, here “helping the experimenter,” while seeing the “nonsharing, punished” tape which also increased sharing in the first experiment, will not increase another prosocial behavior, helping. Each subject was shown one of three tapes used in Experiment I: baseline, “nononly,” and then was asked to help sharing, punished,” and “punishment the experimenter by performing a square-sorting task for 5 min. Number of squares sorted was taken as a measure of how much each subject. helped the experimenter.

232

MORRIS,

MARSHALL.

Ah’D

MILLER

METHOD

Subjects Seventy-eight first, serond, and third grade girls from a fourth local elementary school, also located in a small town predominantly lower middle class, participated in Experiment II. The mean age of these subjects was 8 years, 0 mo. The data from two of the subjects were discarded: one when it was observed that she had a partial paralysis of one wrist which interfered with performing the experimental task; the other when it was discovered that she had served as a subject in Experiment I while attending another school. The remaining 76 subjects were assigned to the three experimental conditions, 25 to each of two and 26 to the third. Equipment

and Apparatus

The experiment was conducted in the children’s own school in a room equipped with a child’s desk, a large table, and several chairs. The same TV camera, TV monitor, and videorecorder were used as in Experiment I. Modeling

Conditions

Three of the tapes which were used in Experiment I were used in Experiment II: a baseline tape, a “nonsharing, punished” tape, and a “punishment only” tape. Because no differences were found in Experiment I for socializing agents, only one set of tapes was used in Experiment II. The experimenter in Experiment II was a 24-year-old male graduate student easily distinguishable from the female socializing agent on the tapes.

As in Experiment I children were brought individually to the experimental room and were assigned to one of the three experimental conditions by order of their arrival. The subject and experimenter again sat at a small desk in front of the TV camera and monitor. The experimenter asked the subject the same general questions as in Experiment I and called her attention to the television monitor with her image on it. After a few seconds the experimenter announced that he had to leave the room for a while, and that while he was gone he wanted the subject to watch another girl on TV. He then left, and the experimental assistant presented one of the video-taped modeling conditions to the child. This time as soon as the tape was finished, the experimental assistant turned the TV camera back on! so that the child once again could see

VICARIOUS

PIYNISHMENT

233

herself on TV. The camera remained on for the rest of thc~ c~xperiment~xl session. The experimental assistant called the exl)c~rimrntc~r back to the room. The experimenter returned and announced that hc was back but that he was going to have to go out again, and told the subject that this time he had something he’d like the subject to help him with Ivhilc he was gone. He placed on the subject’s desk a box containing several thousand lvz in. construction-paper squares of three different colors: red, green, and blue. The experimenter explained that the squares were all mixed up and that he needed to have them separated. He presented the subject with three smaller boxes each of which had eight squares of one of the colors stapled to it, and explained that hc needed to have all the red squares placed in the red box, etc. He told the subject hc would be gone for 5 min and asked her to help him by sorting as many of the squares as she could while he was gone. Hc then left the room for 5 min. The experimental assistant remained in the room throughout the entire session. She was seated partially hidden behind the video-tape equipment, and did not look at or speak to the subjects. The subjects appeared to ignore her almost completely; only one subject spoke to the experimental assistant at any time during the experimental sessions. At the end of the 5 min sorting period, the experimenter returned and told t.he subject it was time to return to her classroom. While he was walking her there, he asked each child to promise not to tell what she had done in t’he experimental room. Informal questions posed to some subjects on the way to the experimental session suggested that the promises had been kept. While the experimenter was taking one subject to the classroom and returning with the next, the experimental assistant prepared the video-tape equipment for the next session and marked and put away the squares sorted by the previous subject. Each child spent approximately 9 min in the experimental room. RESULTS

Mean number of squares sorted as a function of modeling condition and grade are reported in Table 2. A two-way analysis of variance (3 X 31, modeling conditions X grades with unequal cell frequencies, was performed on these data. The analysis of variance revealed a significant main effect for modeling conditions (F(2,67) = 3.41, p < .05). One-tailed t-tests indicated that the “punishment only” mean was significantly greater than the baseline mean (t(49) = 1.80, p < .05) and the “nonsharing, punished” mean (t(49) = 1.62, p < ,055). The “nonsharing, punished,” and baseline means did not differ significantly (t(48) = 0.52).

234

MORRIS,

MARSHALL,

TABLE

AND

MILLER

2

Modeling

Grade

Baseline

Non-sharing, punished

1st 2nd 3rd

178.1 191.8 207.6

175.7 209 .3 223.9

condition Punishment only 185.0 231.1 282.7

The analysis of variance also revealed a significant grades effect (F(2,67) = 6.62, p < .Ol). Comparisons revealed that the first grade mean was significantly lower than both the second and third grade means (t(53) = 2.05, ?, < .05; t(55) = 3.59, p < .Ol, respectively). The difference between the second and third grade means approached significance (t(38) = 1.59; p < .12). The interaction between grade and modeling condition did not approach significance. DISCUSSION

Experiment II further supports the hypothesis that the viewing of noncontingent punishment to a model results in a generalized inhibition against antisocial behaviors. At the same time, the data suggest that contingent vicarious punishment has an effect specific to the punished behavior. One may infer from these findings that in the “nonsharing, punished” condition of Experiment I and the vicarious punishment conditions of previous experiment,s (e.g., Bandura, 1965; Walters, Leat, & Mezei, 1963; Walters, Parke, & Cane, 1965) children learned a specific lesson, namely, that a particular behavior may lead to punishment. As a result, t.hey tended to avoid emitting those particular behaviors. When the vicarious punishment is not attached to specific behaviors, however, the child tends to avoid those behaviors which have been associated with punishment in the past. The finding that punishment alone leads to generalized inhibition of antisocial behaviors is not easily incorporated into existing theoretical statements of the effects of vicarious reinforcement on behavior. Such statements were developed, after all, to account for increases or decreases in observers’ matching or imitative behaviors. Thus they depend on there being specific modeled responses, and are difficult to extend to the noncontingent punishment case where no such specific responses occur. Bandura, however: has dea,lt with the possibility of vicarious re-

VICARIOUS

235

PUNISHMENT

inforcement conveying information about the social environment more general than the specific response contingencies involved. He speculates, for example, that the timing of vicarious reinforcement relative to the model’s behavior may convey information about the punitiveness of the socializing agent: “It also seems highly probable t’hat a person who reprimands a child even before a deviant act actually occurs would be viewed as considerably more forbidding and punitive” (1969, p. 303). This is the sort of information that may have been conveyed by the “punishment only” tape. One could suppose, for example, that a noncontingently punitive socializing agent is perceived as more punitive than a contingently punitive socializing agent, and that this difference accounts, in part, for the findings of Experiment II. Or one could argue more generally that the “punishment only” tape suggests to the child that her present environment is a potentially dangerous one. Either way it would appear that noncontingent vicarious punishment has some informative function. In sum, WC would suggest that noncontingent vicarious punishment conveys to the observer the information that her present environment is a threatening one, and that such information arouses anxiety in her. Such vicarious punishment does not, however, provide any information about what response the observer must avoid in order to prevent punishment. and eliminate her anxiety. A reasonable “strategy” on the part of the child is to perform exclusively those responses which she has previously learned lead to no punishment,. In most children’s learning history, the class of behaviors termed “prosocial” is likely t,o have such a property. REFERENCES B~XDURA.

imitative 589-595.

A. Influence of models’ responses. Journal

reinforcement of Personality

contingencies and Social

on the acquisition Psychology, 1965,

of 1,

A. Principles of Behavior Modification. New York: Holt, Rinehart and 1969. RAND~.A. 8. Vicarious and self-reinforcement processes. In R. Glaser (Ed.), The natzcre of reinforcement. New York: Academic Press, 1971. BERKOWITZ, 1,. Beyond exchange: Ideals and other factors affecting helping and altruism. Unpublished manuscript, University of Wisconsin, 1968. BERKOWITZ, I,.. $ I)ANIEI.S. 1,. Affecting the salience of the social responsibi1it.v norm: Effects of past, help on the response to dependenc.v relationships. JolLrnal of Abnormal and Social Psychology, 1964, 68, 275-281. BOE, E. E.. & CHURCH, R. M. (Eds.) Punishment. Issues and experiments. New York: Appleton-Century-Crofts. 1968. RONEAU, C. A. The effects of violations of assumpt,ions underlying the t-test. Psychological Bulletin. 1960, 57, 49-64.

B.4NDURA,

Winston,

236

MORRIS,

MARSHALL,

AND

MILLER

BRYAN, J. H., & WALBEIC, N. The impact of words and deeds concerning altruism upon children. Child Development, 1970, 41, 747-757. CAMPBELL, B. A., & CHURCH, R. M. (Eds.) Punishment and Aversive Behavior. New I-ok: Appleton-Century-Crofts, 1969. ELLIOTT, R., 6; VASTA, R. The modeling of sharing: Effects associated with vicarious reinforcement, symbolizations, age, and generalization. Jowwrl of Experimental Child Psychology, 1970, 10, 8-15. HANDLON, B. J., & GROSS, P. Thr development of sharing behavior. JOICI,UC~/ oj Bhnomud a/d Socird Psychology, 1959, 59, 425-428. HARRIS, M. B. Reciprocity and gemrosily: Some determinnnts of sharing in children. Child Drcelopment, IOiO, 41, 313-328. IJNDQT-IRT, IX. F. Ilesign and analysis of c.rpe,irtwt2ls irk psychology (/bf/ rrlfcc/rlio0. lioston : Houghton Mifflin, 1956. MARSIIALL, H. M. The effect, of vicarious punishment on sharing behavior in chiltlrcn. Unpublished doctoral dissertation, Dartmouth College, 1972. MIDLARSICY, E., & BRYAN. J. H. Training chnrit,v in children. Jocc~noi cd Pexso~~nlily and Social Psychology, 1967. 5, 429-434 PARICII. R. II. Effwiiveness of punishment as an interaction of intensity. timing. :tgenl nurt.rnance, and cogniii\,c structuring. Chitd De~~elopnLenl. 1969, 40, 213236. ROSEXHAN, D., B WHITE. G. M. ObscrJ-ation and l,ehrarsal as determinants of Iwosocial brha\-ior. fortrnal of P~rso&ity and Social Psychology, 1967, 5, 422-431. WALTERS, R. H.. 6: ANDRE;~. D. Punishment procedures and self-control. Paper rend at the Annual Meeting of the American Psychological Assoriation. Washington, IX. C., 1967. WALTERS~ R. H.. LEAT, M.. & MEZEI, L. Inhibition and disinhibition of response through empnthrtic learning. Canadian Jownal of Psychology, 1963, 17, 23b243. \VALTERS, R. H.. PARKE, 1~. II.. & CASE, V. A. Timing of punishment and the ohscrvntion of ronsequenrcs to othcra as determinants of response inhibition. Journal of Esperimetttrrl Chilcl Psychology, 1965. 2, 10-30.