奈良教育大学学術リポジトリNEAR
The Effects of Verbal Reinforcement
Combinations upon Marble ‑Dropping Performance of Reflective and Impulsive Children
著者 TAMASE Koji
journal or
publication title
奈良教育大学紀要. 人文・社会科学
volume 36
number 1
page range 163‑171
year 1987‑11‑25
URL http://hdl.handle.net/10105/2072
The Effects of Verbal Reinforcement Combinations upon Marble ‑Dropping Performance of Reflective
and Impulsive Children*
Koji Tamase
(Department of Psychology, Nara University of Education, Nara 630, Japan) (Received April 7,1987)
The relationship between verbal reinforcement combinations and children s reflection‑impulsivity was examined. From 145 kindergarten children 30 reflective and 30 impulsive children were selected using Kagan s MFF test. They were given a two‑hole marble‑dropping task under one of the two conditions. For the RN‑NW group the subject was given "Right only after his correct response in the first half of the acquisition period and given "Wrong only after his incorrect response in the second half. For the NW‑RN group the subject was given Wrong only
IIafter his incorrect response in the first half of the acquisition period and given
"Right only after his correct response in the second half. The reflective children responded more slowly and corr∝tly under NW than under RN, while the impulsive children responded similarly through the acquisition period under both RN and NW. It was concluded that reflective children were more sensitive to differential verbal reinforcements than impulsive children.
A number of researchers have examined the effect of verba王reinforcement combinations
upon human learning (Barringer & Gholson,1979; Marshall,1965 for review). They have usually compared the following three verbal reinforcement combinations: (a) "Right after a coir∝t response and "Wrong after an incorr∝t response (RW) , (b) "Right after a correct response and nothing after an incorrect response (RN) , and (c) nothing after a correct response and "Wrong' after an incorrect response (NW). Previous studies have shown, in general, that RW and NW produce equivalent performance which is superior to that produced by RN.
Although most of earlier researchers used a single measure of acquisition to determine the relative effectiveness of verbal reinforcement combinations, Hamilton (1969) used two measures to assess components of RN and NW in a two‑hole marble dropping task. He found that the number of correct responses during the reinforcement period was significantly
greater under NW than under RN, and that the total number of responses under NW did
The author is indebted to Mr. Shigeyoshi Honnouchi for his assistance in collecting and analyzing the data and to Dr. Takeshi Sugimura and Ms. Susan Brady for their helpful comments on
an earlier draft of this paper.
163
164
Koji Tamase
not differ from that under RN. He assumed that number of correct responses reflected an informational asp∝t of verbal reinforcements and that the total number of r鴎ponses refl∝ted a motivational asp∝t.
Tamase (1984) , using a two‑hole marble‑dropping task for kindergarten‑children, performed a similar experiment. His subjects were verbally reinforced under one of the four experimental conditions: "Right‑Nothing through the acquisition period (RN‑RN) ,
'Nothing‑Wrong through the acquisition period (NW‑NW), RN in the first half of the acquisition period and NW in the second half (RN‑NW), and NW in the first half of the acquisition period and RN in the second half (NW‑RN). The correct response was determined as dropping a marble into the hole which was much less preferred by the subject in the operant period. The total number of responses during the reinforcement period was also examined. In the first half of the acquisition period the ratio of corr∝t responses was greater for the NW‑NW and NW‑RN groups than for the RN‑NW and RN‑RN groups. For the RN‑NW group the ratio of correct responses increased from the first half of the acquisition period (RN) to the second half (NW) and approached the NW‑NW group. For the NW‑RN group the ratio of correct responses did not increase in the second half of the acquisition period. The total number of responses in the first half of the acquisition period for the RN‑RN and RN‑NW groups did not differ from that for the NW‑NW and NW‑RN groups, but it decreased early in the second half for the RN‑NW group. Such decrement in the second half of the acquisition period did not occur for the NW‑RN group. These findings suggest that introducing NW, as compared with RN, increases the correct responses while it produces the temporal decrement of the response speed. The author explained these findings as showing that W produced more attention to the task than R and consequently provided more information regarding correctness of the response.
The phenomenon that NW produces a higher proportion of correct r塔ponses and decreases response speed more that RN is similar to the response tendency of reflective and impulsive children (Kagan, Rosman, Day, Albert, & Phillips, 1964). According to Kagan (1965), reflective children take a long time before choosing an answer to a problem and make relatively few errors, while impulsive children choose very quickly without reflecting on the validity of their responses and make many errors.
As for thCjse individual differences many theoretical considerations were proposed (Kogan,
1983; Messer,1976). Kagan and Kogan (1970) has posited that impulsivity may be a function
of high anxiety about response and low anxiety about response accuracy, while reflection
may be a function of high anxiety about response accuracy and low anxiety about response
speed. If this is the case, reflective children in a two‑hole mardle‑dropping task may look
upon verbal reinforcements as given for correctness of their responses and thus become more
attentive to corr∝tness of response. According to Tamase (1984) , this trend will be greater
under NW than under RN condition. On the contrary impulsive children may look upon
verbal reinforcements as given for their response speed and thus try to respond faster when
verbal reinforcement is given. Whether such trend differs in RN and NW conditions or not
is unclear at present.
Several researchers examined the interactive relationship between the two phenomena.
Nuessle (1972) , usingr a concept identification task, found that the response latency following W ("Wrongつwas longer than following C ("Correctつand that the reflective children had a significantly larger difference between W and C latencies than did the impulsive children. Hemry (1973) examined more directly the relationship between verbal reinforcement combinations (RW, RN, and NW) and reflection‑impulsivity.
Although reflective children showed better performance than impulsive children and RN produced poorer performance than RW and NW, no significant interaction of reflection‑impulsivity with reinforcement combinations was obtained. Massan and Schack (1972) , using a two choice discrimination learning task, examined the effect of reinforcement schedules in a verbal reinforcement combinations. Children as the
subjects were assigned either to a positive reinfocement condition (70% R and 30%
W) or a negative reinforcement condition (30% R and 70% W). Reflective children
were expected to perform better than impulsive children, and impulsive children in the negative reinforcement condition would outperform their impulsive counterparts in the positive reinforcement condition. The results indicated that reflective children made more correct responses than impulsive children and that children in the negative condition responded correctly more often than those in the positive condition. However, no significant interactive effect was found.
As the above mentioned studies were rather different from each other in their results, further researches are needed to clarify the relationship between verbal reinforcement combinations and reflection‑impulsivity. The present study was designed to examine the relationship between verbal reinforcement combinations and reflection‑impulsivity in preschool children. It was predicted that the tendency that NW produces a higher proportion of correct responses than RN will be shown more markedly for reflective children than for impulsive children. It also was predicted tentatively that the tendency for RN to produce faster response speed than NW will be shown more markedly for impulsive children than for reflective children.
Method
Experimental Design
A 2 (Verbal Reinforcement Combination) × 2 (Reflection‑Impulsivity) factorial design was used. The first factor was the condition of verbal reinforcement combination (RN‑NW versus NW‑RN) and the second factor was subjects cognitive style (Reflective versus Impulsive).
Selection of the Subjects
The subjects were chosen from a population of 145 kindergarten children (68
166
Koji Tamase
boys and 77 girls) attending a public kindergarten in Nara, on the basis of their performance on Kagan s Matching Familiar Figures Test (MFF, Kagan et al.,1964).
The MFF consisted of 12 items, each item containing a standard stimulus and six variants. The subject s task was to choose the one variant that exactly matched the standard stimulus. Usually when the subject is incorrect the child is allowed to continue until one correct or 6 incorrect choices are made. In this experiment, however, the subject was allowed only one choice whether it is correct or incorrect. This method was used by Ward (1968) in a modified form of MFF. This modification of the procedure was used to shorten the time spent on one subject. Following each choice the subject was given verbal reinforcement (right or wrong). The total number of errors and latency to the first response on each item were recorded for all subjects.
The mean age of the population was5 yr. 7 mO. (5:1 ‑6:1).
The median response latency for total subjects was 9.8 sec. and the median number of errors was 8.0. 34 subjects who scored above the median on response latency and below the median on errors were classified as the reflective children. 37 subjects who scored below the median on response latency and above the median on errors were classified as the impulsive children. For the reflective children the mean response latency was 24.6 sec. (SD ‑ 21.5) and the mean number of errors was 5.4 (SD
‑ 1.7), and for the impulsive children they were 6.2 sec.(SD ‑ 2.3) and 9.7 (SD
‑ 0.8), respectively. Among the selected subjects, those who responded to only one hole of the two in the operant period of the marble‑dropping task were excluded as having the response preference. Eventually, 60 subjects were assigned to either reflective RN‑NW (MFF latency: 26.0, error: 5.8), reflective NW‑RN (17.6, 5.3), impulsive RN‑NW (6.2, 9.7), or impulsive NW‑RN (6.5, 9.8). 15 subjects were involved in each cell. Their mean age in each group was 5 yr. 8 mO. (5:3 ‑ 6:2).
Materials
The task was to drop a marble into one of 7.5 × 23 × 16.5 cm size. Each hole was were 5 cm apart. Each hole was edged with edged in light blue and the other pink. 200 diameter) were prepared, and a container for
of the two holes in a pasteboard box 1.7 cm in diameter and the two holes color of 1.1 cm width. One hole was transparent blue marbles (1.2 cm in the marble was located on the desk.
Procedure
The experiment was done individually in a room in the kindergarten one month after the execution of the MFF test. The escorted subject was seated at the desk on which the box for marble‑dropping was located, and faced the experimenter who was seated at the other side of the desk. The subject was given the following
instructions :"Now you are going to play a marble‑dropping game. Here is a box with
two holes. There are many marbles in the container. Pick up a marble and drop it into one of the holes. You can drop the marble into either one of the two holes.
Please keep dropping them until I say stop.
The subject was observed for 1 minute to determine which hole of the two he preferred to drop. During this period verbal reinforcement was never given to his responses. Following the operant period the experimenter began to give "Right for the subject's choice of the nonpreferred hole or give "Wrong for the subject's choice of the preferred hole according to the experimental conditions. The reinforcement period was 4 minutes. Half of the reflective and the impulsive subjects were given
"Right after a correct response (a choice of the nonpreferred hole) and nothing after an incorrect response (a choice of the preferred hole) in the first half of the reinforcement period and they were given "Wrong after an incorrect response and nothing after a correct response in the second half (RN‑NW). The remaining half of the reflective and the impulsive subjects were given "Wrong after an incorrect response and nothing after a correct response in the first half of the reinforcement period and they were given "Right after a correct response and nothing after an incorrect response in the second half (NW‑RN主
Results
The ratio of correct responses
The ratio of correct responses was calculated for each block of 1 minute for the individual subject. The mean ratios are shown in Table 1. For the scores of the operant period no significant difference was found among the groups. A 2 (Reflective
versus Impulsive) × 2 (RN‑NW versus NW‑RN) × 5 (Trial Block) analysis of varianceon the mean ratios was done. The main effect of Trial Block (F ‑ 4.73, df ‑ 4, 224, P <.Ol, MSe ‑ 0.03), the interaction of Reinforcement Combination by Trial
Table 1
Mean ratio of correct respollses for reflective and impulsive children under RN‑NW and NW‑RN reinforcement conditions
Acquisition blocks of one minute Operant Before shift After shift
1 2 3 4
Reflective
RN‑NW .39 .48 .47 .62 NW‑RN .43 .64 .77 .59 Impulsive
RN‑NW .39 .55 .58 .60
NW‑RN .45 .60 .64 .62
168
Kou Tamase
Block (F ‑ 5.31, df‑ 4, 224, P <.01, MSe ‑ 0.05) and the interaction of Reflection‑Impulsivity by Reinforcement Combination by Trial Block (F ‑ 3.24, df
‑ 4, 224, Pく.05, MSe ‑ 0.03) were significant. The other variances were not significant.
The significant interaction of Reinforcement Combination by Trial Block shows that the performance was better under NW than RN in the acquisition 2 (/ ‑ 3.12, df ‑ 280, P <‑01). Further tests of the significant second‑order interaction revealed the following results. For the reflective children in NW‑RN group showed better performance than the RN‑NW group in the acquisition 1 (t ‑ 1.96, df ‑ 280, P
<.O5) and 2 (t ‑ 3.67, df ‑ 280, P <‑01), while the NW‑RN group showed worse performance than the RN‑NW group in the acquisition 4 (t ‑ 2.20, df ‑ 280, P
¥.O5). For the impulsive children significant difference between the RN‑NW and the NW‑RN groups was not found in any block. These results show that the reflective children respond more correctly under the NW condition than under the RN condition.
The total number of responses per block
The mean total number of responses per block for the four groups is shown in Table 2. A 2 × 2 × 5 analysis of variance revealed that the main effect of trial Block (F ‑ 15.25, df‑ 4, 224, P<.Ol, MSe ‑ 6.90) and theinteraction of
Reinforcement Combination by Trial Block (F ‑ 3.47, df ‑ A, 224, Pく.01, MSe
‑ 38.61) were significant. The other variances were not significant. The significant interaction indicates that the RN‑NW groups were superior to the NW‑RN groups in the acquisition 1, 2, and4 (/ ‑ 2.35, 1.99, and2.12in order, df‑ 280, P く.05).
Table 2
Mean total number of responses per block for reflective and impulsive children under RN‑NW and NW‑RN reinforcement conditions
Acquisition blocks of one minute Operant Before shift After shift
1 2 3 4
Reflective
RN‑NW 27.00 28.60 30. 73 29.07 30.93 NW‑RN 24.93 22.13 25.20 25.27 27.00 Impulsive
RN‑NW 25.13 25.60 27.73 25.87 29.27
NW‑RN 26.00 24.53 26.87 24.87 26.40
To examine whether the introduction of verbal reinforcement produced some changes in the response or not, the change scores of the total number of responses per block from the operant period to the acquisition 1 and from the acquisition 2 to 3 were calculated. The mean change scores of the Reflective and the Impulsive groups in each reinforcement condition are shown in Table 3. A 2 × 2 analysis of variance was done on the change scores in the first introduction of verbal reinforcements. A significant main effect of Reinforcement Combination was obtained (F ‑ 8.85, df‑ 1, 56, P <.Ol, MSe ‑ 16.99), which showed that NW decreased the subject s response speed while RN rather increased it. A similar analysis of variance was done on the scores in the second introduction of verbal reinforcement (the shift of reinforcement), but no significant results were obtained.
Tab e 3
Mean change scores of the total number of responses per block for reflective and impulsive children under RN‑NW and NW‑RN
reinforcement conditions and the number of subjects in each direction of change in each group
AcquisitionトOperant Acquisition 3‑2
# of Ss
Mean Mean
+ 0 ‑
# of 5s + 0
Reflective
RN‑NW NW‑RN Impulsive
RN‑NW NW‑RN
1.60 ‑1.66
‑2.80 1 9 0.07
0.47 ‑1.86
‑1.48 0 10 ‑2.00
Discussion
The performance differences in reflective and impulsive children under different verbal reinforcement combinations were examined using the ratio of correct responses and the total number of responses per block in the two‑hole marble‑dropping task.
As for the ratio of correct responses the reflective children showed better
performance under NW condition than under RN condition. That is, the reflective
NW‑RN group showed significantly better performance than the reflective RN‑NW
group in the first half of the acquisition period, while the reflective NW‑RN group
showed worse performance than the reflective RN‑NW group in the second half of
the acquisition period. On the other hand, the impulsive children showed similar
performances through the acquisition period under both RN and NW conditions. Thus
170 Koji Tamase
the performance differences under RN and NW conditions were greater for reflective children than for impulsive children. According to Hamilton (1969) the ratio of correct responses is assumed to reflect the informational aspect of verbal reinforcements.
From this point of view the present results can be interpreted as showing that the reflective children got more information under NW condition that under RN condition, while the impulsive children got about equivalent information under RN and NW conditions.
The findings for the ratio of correct responses also may be interpreted by assuming that reflective children have more differential sensitivity to verbal reinforcements than impulsive children. When reflective children who are predisposed to respond carefully were given "Wrong, they might attend to their responses much more than before. As a consequence, they might get more information from the NW reinforcement combination. On the other hand, when reflective children were given
"Right, they might attend to their responses much less than before. Thus they might get less information from the RN reinforcement combination.
As for the total number of responses per block no significant second‑order interaction was found. Thus the prediction that the tendency for RN to produce faster response speed than NW will be shown more markedly for impulsive childern than for reflective children was not confirmed in the present study. The change scores of the total number of responses (Table 3) showed that regardless of children s reflection‑impulsivity NW decreased the response speed in the first introduction of reinforcements. According to Hamilton (1969) the total number of responses per block is assumed to reflect the motivational aspect of verbal reinforcements. From this point of view the present results can be interpreted as showing that NW weakened children s motivation for performing the task.
It is interesting to note the relationship between the ratio of correct responses and the total number of responses in reflective children. In the reflective NW‑RN
group the ratio of correct responses increased markedly in the first half of the acquisition period and decreased in the second half, while the total number of responses decreased in the first half of the acquisition period and recovered slowly in the second half. In the reflective RN‑NW group, on the other hand, the ratio of correct responses did not increases so much in the first half of the acquisition period and increased markedly in the second half, while the total number of responses increased in the first half of the acquisition period and decreased temporarily in the second half. The impulsive groups did not show such a differential t・rend of responding in response to differential verbal reinforcements. An important finding is that reflective children are much more sensitive to differetial verbal reinforcements than impulsive children.
When they were given positive verbal reinforcements, they were likely to respond fast
and incautiously, and when they were given negative verbal reinforcements (verbal
punishments), they were likely to respond slowly and cautionsly.
References
Barnnger, C.,& Gholson, B. 1979 Effects of type and combination of feedback upon conceptual learning by children: Implications for research in academic learning. Review of Educational Research, 49, 459‑478.
Hamilton, M,L. 1969 Reward and punishment in child discrimination learning. Developmental Psychology, 1,735‑738.
Hemry, F. P. 1973 Effect of reinforcement conditions on a discrimination learning task for impulsive versus reflective children. Child Development, 44,657‑660.
Kagan, J. 1965 Impulsive and reflective children: Significance of conceptual tempo. In J.D. Krumboltz (Ed.). Learning and the educational process. Chicago: Rand McNally,
Pp.133‑161.
Kagan, J., & Kogan, N. 1970 Individual variation in cognitive processes. In P.H.Mussen (Ld.), Carmichael's manual of child psychology,(vol.1)(3rd ed.). New York: Wiley,Pp.
1273‑1365.
Kagan, J.,Rosman, B‑ L.,Day, D., Albert, J.,& Phillips, W. 1964 Information processing in the child: Significance of analytic and reflective attitudes. Psychological Monographs, 78(1, Whole No.578 .
Kogan, N. 1983 Stylistic variation in childhood and adolescence. In P.H.Mussen (Ed.), Handbook of child psychology (Vol. 3)(4th ed.). New York : Wiley, Pp. 672‑706.
Marshall, H. 1965 The effect of punishment on children: A review of the literature and a suggested hypothesis. Journal of Genetic psychology, 106, 23‑33.
Massari, D‑ J., & Schack, M. L. 1972 Discrimination learning by reflective and impulsivel children as a function of reinforcement schedule. Developmental Psychology, 6, 183.
Messer, S. B. 1976 Reflection‑impulsivity: A review. Psychological Bulletin, 83, 1026‑1052.
Nuessle, W. 1972 Reflectivity as an influence on focusing behavior of children. Journal of Experimental Child Psychology, 14, 256‑276.
Tamase, K. 1984 The function of "Wrong in verbal reinforcement combinations. Japanese Journal of Psychology, 54, 386‑389. (In Japanese with English Summary)
Ward, W. C 1968 Reflection‑impulsivity in kindergarten children. Child Development,^,
867‑874.
