You are on page 1of 7

THE EFFECTS OF REWARD AND PUNISHMENT UPON CHILDREN'S ATTENTION, MOTIVATION, AND DISCRIMINATION LEARNING

KENNETH L. W i n E and EUGENE E. GROSSMAN

University of Arkansas WiTTE, KENNETH L., and GBOSSMAN, EUGENE E . The Effects of Reward and Punishment upon Children's Attention, Motivation, and Discrimination Learning. CHILD DEVELOPMENT, 1971, 42, 537-542. The performances of 60 kindergarten-aged children on a tactile form discrimination task were compared under S reinforcement conditions {reward only, punishment only, reward and punishment). Reward and punishment were nonverbal, consisting of tokens which were exchanged for a toy, and a 98 db tone. The numbers of correct responses for the 2 punishment groups did not differ; both groups made more correct responses than the reward only group. Additional data suggested that the preceding performance differences were due to attentional, rather than motivational, differences among the 3 groups.

Over the past decade a number of studies have examined the effects of nonverbal reward and punishment upon children's learning. The results of these studies have usually indicated that punishment, either alone or in combination with reward, leads to better performance than does reward alone. While thefindingsare fairly consistent, the process by which punishment serves to enhance discrimination performance is currently unresolved. Two hypotheses offered have involved an attentional mechanism. Penney (1967) and Spence and Dunton (1967) have suggested that presentation of the reward may serve to distract the child's attention and thus retard learning, while others (Penney 1967; Stevenson,
Thanks are due to Mrs. Irene Bint and Mrs. Gloria Bishop, head teachers, for their cooperation in providing Ss, and to Mr. Kenneth Johnson for his technical assistance. Authors' address: Psychology Department, University of Arkansas, Payetteville, Arkansas 72701.
ICMld Deidopment, 1971, 42, 537-542. 1971 by the Society for Besearch in Child Development, Inc. All rights reserved.l

CHILD DEVELOPMENT

Weir, & Zigler 1959) have indicated that punishment may increase the child's attention and thus facihtate learning. A third hypothesis, motivational in nature, suggests that punishment serves to facilitate learning by increasing the child's motivational level (Brackbill & O'Hara 1958). In the above studies, the dependent variable has been either number of trials to a performance criterion or number of correct responses in a fixed number of trials. With one exception (Penney 1967), no one has attempted to measure possible changes in either the child's attentional or motivational state. Penney presented children a tactile discrimination problem and found that group P (punishment only) performed better (i.e., fewer trials to criterion) than group R & P (reward and punishment) which, in turn, performed better than group R (reward only). The child's attention was assessed by the number of orienting responses occurring per trial, that is, the number of times the child touched the stimuli. The greatest number of orienting responses was made by group P, while group R made the fewest. Penney therefore concluded that reward and punishment influence discrimination learning via an attentional mechanism. While the preceding study provides evidence for an attentional hypothesis, it does not preclude the possibility that a concomitant increase in the child's motivation also results from punishment. It would thus seem important that further research employs response measures to assess possible motivational changes, as well as response measures to assess possible attentional changes. The purpose of the present study was to test the hypothesis that punishment may serve to make the child more motivated as well as more attentive. To accomplish this aim, the type of discrimination problem employed by Penney was used, and, in addition, a means of measuring the child's motivational level was included. This was achieved by presenting the stimuli upon scale platforms and then measuring the force with which Ss made their choice responses. This type of response measure has been successfully employed in studies concerned with the possible motivating effect of frustration upon children's behavior (e.g., Holton 1961).
METHOD

Subjects.The Ss were 60 children from two private kindergartens located in Fayetteville, Arkansas. Ten male and 10 female Ss were randomly assigned to each of three treatment groups: group R, group R & P, and group P. Apparatus.The discriminative stimuli consisted of two white wooden blocks (a circle 1.5 inches in diameter and an equilateral triangle with 1.5-inch sides). On each trial the stimuli were placed on the platforms of two 25-lb capacity scales, modified so that the weight indicator
538

KENNETH L. WITTE A N D EUGENE E. GROSSMAN

remained at the dial location signifying maximum depression until released by E. The scales were positioned 10 inches from each other and were contained within a gray wooden box. The front of the box had a horizontal opening parallel to, and at the same height as, the scale platforms located inside. A white curtain with an elasticized slit across its length covered the opening. A small drawer was located at the base of the box directly beneath the curtained opening. A 98-db, 2,800-cps tone, used for the punishment conditions, emanated from two speakers placed one on each side of the S. The duration of the tone was 3 seconds. Procedure.All Ss received 36 trials on a two-choice tactile form discrimination problem. A noncorrection procedure was used with a Fellows' (1967) series employed to determine the spatial location of the stimuli on each tri^l. For one half the Ss in each treatment group the circle served as the positive stimulus, whereas the triangle was the positive stimulus for tl^e remaining Ss. Each S wfts instructed to reach through the opening in the curtain, feel the two blocks, and then push down on the block he thought to be correct. All Ss in groups R and R & P were also told that each time they made a correct response the drawer on the face of the apparatus would open presenting a token which they could remove and that if they won enough tokens they could exchange them for a small toy. All Ss in groups P and R & P were instructed that each time they made an error they would hear a loud noise. The noise was then demonstrated and the Ss in group P were further advised that if they performed "well" they would receive a toy.
RESULTS A N D DISCUSSION

Three analyses of variance were conducted, one for each response measure. Each analysis involved a Lindquist (1953, pp. 267 ff.) Type I design in which the between-Ss' variable was treatment groups while the within-Ss' variable was trial blocks (three blocks of 12 trials per block). Tests for homogeneity of variance using F max indicated that this assumption was not met for the correct response and orienting response data but was met for the response force data. Therefore, a .025 significance level was used for all statistical tests involving the first two measures, while a .05 significance level was used for the response force data. The correct response data are presented in table 1. The main effect for treatment groups was significant {F = 21.28; df = 2/57), as was the main effect for trial blocks {F = 92.65; df = 2/114). The interaction between treatment groups and trial blocks was also significant {F = 4.43; df = 4/114). Follow-up tests indicated that (a) groups P and R & P did not differ on any of the three trial blocks, and (b) the numbers of correct
539

CHILD DEVELOPMENT

TABLE 1
MEAN NUMBER OF CORRECT RESPONSES FOR THE THREE TREATMENT GROtrps ACROSS TRIAL BLOCKS TRIAL BLOCKS

One
TREATMENT GROUP

Two SD 1.35 1.53 1.67 M 7.85 10.45 11.25 29.55 SD 2.50 1.96 1.67 M

Three
OVERALL

SD 2.68 1.50 1.02

6.85 Group R . . . Group R& P . . . 7.05 7.90 Group P. Overall M 21.80

10.00 11.40 11.55 32.95

24.70 28.90 30.70

responses for groups P and R & P on trial blocks two and three, but not one, were reliably greater than those for group R. The preceding findings of better discrimination performance for the two punishment groups relative to group R are consistent with those of previous studies which have also used a noise as punishment (Penney 1967; Penney & Lupton 1961; Spence & Dunton 1967; Spence & Segner 1967). In addition, the finding of superior performance for group R & P relative to group R is consistent with the results of studies which have used response cost, that is, loss of a reward object, as the method of punishment (Brackbill & O'Hara 1958; Stevenson et al. 1959; Whitehurst 1969). However, the present findings of superior performance for group P relative to group R, and comparable performance for the two punishment groups, are not consistent with the results of the Whitehurst study in which the performance of group P was comparable to group R and inferior to that for group R & P. Of the three studies which have used response cost as the method of punishment, the Whitehurst study is the only one which has included a punishment-only group. Thus, one direction for future research would be to see if the results obtained by Whitehurst can be replicated, and to directly compare the relative effectiveness of these two modes of punishmentresponse cost versus noxious stimulation. The average numbers of orienting responses per trial were 1.21,1.38, and 1.46 for groups R, R & P, and P, respectively. The main effect for treatment groups was significant {F = 11.07; df = 2/57). The numbers of orienting responses across trial blocks increased significantly (F = 39.52; df = 2/114). The interaction between treatment groups and trial blocks was also significant (F = 7.46; df = 4/114). Follow-up tests indicated that: (a) groups R and R & P differed significantly only on trial block two; (6) groups R and P differed significantly on the first two trial
540

KENNETH L. WITTE AND EUGENE E. GROSSMAN

blocks but not the third; and (c) groups R & P and P differed significantly only on trial block one. The preceding findings are fairly consistent with those obtained by Penney (1967), and they can be viewed as supporting an attentional hypothesis, inasmuch as Ss in the two punishment groups were more attentive, that is, made more orienting responses, and more correct responses, than Ss in group R. Whether this difference was due to the fact that punishment increased the attentional level of the Ss in the two punishment groups or because presentation of the reward served to distract the Ss in group R is not clear. Nonetheless, E's observations that Ss in the two punishment groups, particularly group P, tended to remain oriented toward the apparatus during the intertrial interval, whereas Ss in group R often counted and manipulated their tokens, suggest that both of these factors were operating. Inspection of the response force data indicated that response force, on the average, was greatest for group R and lowest for group P. However, the main effect for treatment groups, as well as the main effect for trial blocks and the interaction between these two variables, were all nonsignificant. The failure to obtain any response force differences among the three groups, in cor^junction with the superior discrimination performance of the two punishment groups relative to group R, can be interpreted as being inconsistent with the hypothesis of Brackbill and O'Hara (1958) that punishment facilitates children's discrimination learning by increasing their motivation. Since significant differences were found among the three groups in terms of numbers of orienting responses, it would seem that the most parsimonious explanation of the effects of punishment upon discrimination learning would be an associative (attentional) rather than a motivational interpretation.
REFERENCES
Brackbill, Y., & O'Hara, J. The relative effectiveness of reward and punishment for discrimination learning in children. Journal of Comparative and Physiological Psychology, 1958, 51, 747-751. Fellows, B. J. Chance stimulus sequences for discrimination tasks. Psychological Bulletin, 1967, 67, 87-92. Holton, R. B. Amplitude of an instrumental response following the cessation of reward. Child Development, 1961, 32, 107-116. Lindquist, E. P. Design and analysis of experiments in psychology and education. New York: Houghton Mifflin, 1953. Penney, R. K. The effect of reward and punishment on children's orientation and discrimination learning. Journal of Experimental Psychology, 1967, 75, 140-142. Penney, R. K., & Lupton, A. A. Children's discrimination learning as a function of reward 541

CHILD DEVELOPMENT
and punishment. Journal of Comparative and Physiological Psychology, 1961, 34,

449^51. Spence, J. T., & Dunton, M. C. The influence of verbal and nonverbal reinforcement combinations in the discrimination learning of middle- and lower-class preschool children. Child Development, 1967, 38, 1177-1186. Spence, J. T., & Segner, L. L. Verbal versus nonverbal reinforcement combinations in the discrimination learning of middle- and lower-class children. Child Development, 1967, 38, 29-38. Stevenson, H. W.; Weir, M. W.; & Zigler, E. P. Discrimination learning in children as a function of motive-incentive conditions. Psychological Reports, 1959, 5, 95-98. Whitehurst, G. J. Discrimination learning in children as a function of reinforcement condition, task complexity, and chronological age. Journal of Experimental Child Psychology, 1969, 7, 314-325.

542