奈良教育大学学術リポジトリNEAR
Utilization of Perceptual and Conceptual Attributes in Children's Discrimination Learning as a Function of Age and Category Similarity
著者 SUGIMURA Takeshi, YASUHARA Katsuko journal or
publication title
奈良教育大学紀要. 人文・社会科学
volume 28
number 1
page range 209‑218
year 1979‑11‑15
URL http://hdl.handle.net/10105/2478
奈良教育大学紀要 第28巻 第1号(人文・社会)昭和54年 Bull. Nara Univ. Educ, Vol. 28, No.1 (cult.&soc.), 1979
Utilization of Perceptual and Conceptual Attributes in Children's Discrimination Learning as a
°
Function of Age and Category Similarity
Takeshi SuGIMURA
Department of Psychology, Nara University of Education, Nara, Japan and
Katsuko Yasuhara
(Received April 24, 1979)
288 children from 6 to 10 yr. of age were trained on a redundant discrimination task which could be solved by using either perceptual or conceptual attributes of the instances and they were then given two probe trials to assess which attribute they had used in the discrimination learning. The main results obtained from the probe trials showed that (a) the probability to use perceptual attributes decreased and that to use concepts increased with age, both of which were almost equal at 6 yr. of age, and that (b) the conceptual attributes were more often used in the dissimilar‑category task than in the similar‑category task. It was suggested that use of attributes in discrimination learning depended on children's cognitive development and the nature of the task but not on the degree of learning and the feedback information on Trial 1.
In our previous study (Sugimura & Terao, 1976) children from 3 to 6 yr. of age were given either a conceptual, a half‑conceptual, or a nonconceptual sorting task. The results showed that 3‑yr.‑old children learned the three tasks at almost the same rate and that 6‑
yr.‑old children learned the conceptual task faster and the half‑conceptual task slower than the nonconceptual task, which showed no significant age difference. These were interpreted to show that the rate of learning instance‑response associations remained constant from 3 to 6 yr. of age, whereas children's ability to use concepts in learning the sorting task increased through these ages.
Although colored pictures of instances were used in that study, the perceptual attributes of instances such as color and shape were not manipulated as an experimental variable.
Sugimura and Kunimoto (1976) attempted to assess the effect of the perceptual attributes (shape) of instances on learning the sorting tasks. Children from 3 to 5 yr. of age were given either the conceptual or the half‑conceptual task with instances of fruit and vegetable.
Each task could or could not be solved on the basis of perceptual attributes (roundness and
* The authors are indebted to Mrs. Yoko Sato for her valuable comments and editing the English
phraseology of the manuscript.
209
210
Takeshi SuGIMURAslenderness) of the ii‑stances. The result that 3‑ and 4‑yr.‑old children 、vere more susceptible to the perceptual attributes than 5‑yr・‑old children was in line with developmental theories of cognition and learning (Bruner et al., 1966; Gollin & Saravo, 1971). The general hypothesis proposed by Gollin and Saravo is that there is a shift in cognitive behavior from operations dominated by perceptuaトmotor tendencios to operations dominated by conceptuaレ verbal tendencies.
In a recent study Sugimura (1979) attempted to assess children's use of perceptual and conceptual attributes in learning the sorting task by the use of a probe method. Children were trained on a redundant sorting task which could be solved by using either perceptual or conceptual, or both attributes of the instances, and they were then given two probe trials to assess which attribute they had used in the sorting task. In addition to age factor, category similarity of instances was incorporated as a second variable. The similar‑category task consisted of four instances of fruits and vegetables which had a common superordinate concept (food). The dissimilar‑category task consisted of four instances of fruits and
musical instruments. The resu一ts in the probe trials showed that the probability to use
conceptual attributes increased with age and was more facilitated in the dissimilar‑category task than in the similar‑category task. The age by task interaction was obtained, which showed that developmental changes in the attribute use depended on not only age but also category similarity of instances used in the task.
This study was designed to see whether the results obtained in the sorting task were also shown m a simultaneous discrimination task or not. The sorting tasks used in our previous studies were regarded as a kind of conditional successive discrimination tasks. As simultaneous discrimination tasks had been used in most of the discrimination learn】ng studies, this type
of task was also used in the present study. It ヽvas hypothesized that the probability to use
perceptual attributes would decrease and that to use conceptual attributes would increase with age, and that the conceptual attributes would be more often used in the dissimilar‑
category task than in the similar‑category task. In addition to age and task variables, two factors were incorporated in this study, (a) To see whether degree of discrimination learning would differentially affect the use of attributes or not, the subjects were given two probe trials after they reached two different criteria of discrimination learning, (b) To see the effect of correct and incorrect responses in the first trial of discrimination learning on children's use of attributes, half of the subjects were given a correct feedback to Trial 1 response and the remaining subjects were given an incorrect feedback.
Method
DC∫ign and ∫ubject∫
A 2 (Task: similar vs. dissimilar) ×2 (Feedback on Trial 1: correct vs. incorrect) ×3 (Age: 6, 8, and 10 yr.) factorial design was used. The subjects were 288 children consisting
of 96 kindergartners with a mean age of 6 : 1, 96 2nd一graders with a mean age of 7 : 8,
Utilization of Perceptual and Conceptual Attributes in Children's Discrimination Learning 211
and 96 4th‑graders with a mean age of 10 : 0. Equal numbers of male and female subjects were involved in each age level and were divided into four groups according to whether the subjects were given the similar‑ or the dissimilar‑category task, and whether they were given a correct or an incorrect feedback on Trial 1.
Materials
The similar‑category task consisted of three pairs of instances of fruit and vegetable; an orange and a carrot, and an apple and a cucumber for the discrimination and a banana and a cabbage for the probe. The dissimilar‑category task consisted of three pairs of
. . : 管 . * >
詛・ ft. ・ サ * . * ・ ・
サ . ・ .. ■ ̀ ■ .
、 、 、
、 、 、
、 . 、
、 、
、 、 、
、 、
‑ .
k I
‑. ′ ′ .. ...
′ ′ ′ ′ ..
.′
Discrimination
1
Probe
Fig. 1 An example of similar‑category task
212
Takeshi SuGIMURAinstances of fruit and musical instrument; an orange and a pipe, and an apple and a harmonica for the discrimination and a banana and a tambourin for the probe. Each instance was drawn in black lines on a 4. 5‑×5. 5‑cm white paper and two instances were pasted on a 9‑×17‑cm white card side by side.
Fig. 1 illustrates an example of the similar‑category task. In addition to this sample, three types of pairs were provided with the reverse left‑right position of the arrangements of instances. Four pairs for the discrimination were presented in a random order with the restriction that the same pairs did not appear successively. One of the probe pairs was presented in the丘rst probe trial and the other was presented in the second. As was shown in Fig. 1, round instances (an orange and an apple) of fruit were paired with slender instances (a carrot and a cucumber) of vegetable in the discrimination, and a slender instance (a banana) of fruit was paired with a round instance (a cabbage) of vegetable in the probe.
Similarly, four pairs for the discrimination and two pairs for the probe were provided for the dissimila卜category task. The round instances (an orange and an apple) of fruit were paired with slender instances (a pipe and a harmonica) of musical i去strument in the discrimination, and a slender instance (a banana) of fruit was paired with a round instance (tambounn) of musical instrument in the probe.
Procedure
The subjects were tested individually. After a subject was seated, six cards with each instance to be used in the learning task were presented and the subject was asked to name each instance. If he could not name correctly, the experimenter taught him its correct name. Then the subject was presented a card for the discrimination and was given the following instruction.
Now, let's play a card game. I'm going to show you many cards one at a time.
Each card has two pictures. One of them is always right and the other is always wrong.
Your task is to guess which one is right and to point at it. If you choose the right picture, I will say HHit" but if you choose the wrong picture, I will say HMiss". Try to get as many "Hi f's as possible.
The subject was given the cards for the discrimination one by one at his response rate.
The correct and the incorrect instances in the discrimination were determined according to the subject's response on the丘rst trial. The subjects assigned to the correct feedback condition were given a correct feedback ("Hit") on their 丘rst choices whichever instance they had chosen, and thereafter a subject's chosen instance and the other instance of the same category were determined as correct for this subject. The subjects assigned to the incorrect feedback condition were given an incorrect feedback ("Miss") on their first choices whichever instance they had chosen, and thereafter a subject's unchosen instance and the other instance of the same category were determined as correct for this subject.
The following is an example for the similar‑category task. At the first trial the subject
Utilization of Perceptual and Conceptual Attributes in Children's Discrimination Learning 213
assigned to the correct feedback condition was givenパHit" on either choice of an orange or
a carrot. If the subject selected an orange, an orange and an apple were determined as the correct instances for this subject. If the subject seleted a carrot, a carrot and a cucumber were determined as the correct instances. The subject assigned to the incorrect feedback condition was given "Miss" on his first choice. If the subject selected an orange, a carrot and a cucumber were determined as the correct instances for this subject. If the subject selected a carrot, an orange and an apple were determined as the correct instances.
When all the subjects reached a criterion of four successive correct responses, the first probe pair was presented, where the subjects were given "Hit" whichever instance they had chosen. Immediately after this probe trial, the subjects were again trained on the same discrimination and when they reached a criterion of eight successive correct responses after the first probe the second probe pair was presented. After choosing an instance in the probe pair, the subjects were required to verbalized the reason why they chose the instance.
Based on the choice in each probe trial, the subjects were classified as either perception or concept users. For the similar‑category task, if the subject trained with the round fruits as correct in the discrimination selected a banana in the probe, he was classi丘ed as a concept user and if the subject selected a cabbage, he was classified as a perception user. If the subject trained with the slender vegetables as correct in the discrimination selected a cabbage in the probe, he was classi丘ed as a concept user and if the subject selected a banana, he was classi丘ed as a perception user. For the dissimilar‑category task, if the subject trained with the round fruits as correct in the discrimination selected a banana in the probe, he was classified as a concept user and if the subject selected a tambourin, he was classified as a perception user. If the subject trained with the slender musical instruments as correct in the discrimination selected a tambourin in the probe, he was classi丘ed as a concept user and if the subject selected a banana, he was classified as a perception user.
Results
Discrimination learning
A2 (Task) × 2 (Feedback) × 3 (Age) analysis of variance for the mean trials to reach
four successive correct responses was performed by the use of V夏子すすtransformed scores,
which showed signi丘cant main effects of age (F‑5.62, df‑2/276, P<.01) and feedback
(F‑H.03, df‑1/276, P<.01). The mean trials were 6.2, 4.5, and 3. 1 at 6, 8, and 10 yr.
of age, respectively, and were 3. 1 and 5.6 for the correct and the incorrect feedback conditions, respectively. The obtained feedback effect is interpreted to show that the
feedback information on the subject's first response plays an important role in learning the discrimination task.
Although the age by feedback interaction was not s▲gllificant, tl‑e mean trials in the
correct and the incorrect feedback conditions were 6.0 and 6.4 for 6‑yr.‑olds, 3. 1 and 5. 7
ME!
Takeshi SuGIMURAfor 8‑yr.‑olds, and 1.6 and 4.6 for lO‑yr.‑olds, respectively. These show that children's ability to use the correct feedback increases with age whereas that to use the incorrect feedback remains almost constant through these ages. The results may be interpreted to show that children's learning strategy changes with age.
Probe trial
First trial‑The percentages of perception and concept users were calculated on the first probe trial and two types of statistical analyses were performed. Fig. 2 shows main parts of the obtained results.
ヽ、
・ ‑、 Perception users (Similar) ヽ、、 1
ヽ
\、 Perception users (Dissimilar)
\ ヽ
ヽ
6 yr. 8 yr. 10 yr.
Fig. 2 Performances in the丘rst probe trial
(a) A2 (Task) × 2 (Feedback) × 3 (Age) analysis of variance for the perception users was performed by the use of arcsin transformed scores, which showed signi丘cant main
effects of age (f‑32. 12, df‑2, P<.01) and task (*2‑5.82, df‑¥, P<.05). An analysis of variance for the concept users showed significant main effects of age (^2‑32. 12, df‑2, P<.01) and task (z2‑5.82, df‑¥, P<.05).
(b) A2 (Attribute) × 3 (Age) j^‑test for the similar‑category task was sign捕cant (x2‑8. 88, df‑2, P<. 05), which showed that the percentages of perception users decreased with age but those of concept users increased with age. The percentage of concept users was greater than that of perception users at 8 (P<.05) and 10 (P<C.01) yr. of age.
A2×3 #2‑test for the dissimilar‑category task was also significant (^2‑21. 10, dj‑% P<. 01), which showed that the percentages of perception users decreased with age but those of concept users increased with age. The percentage of concept users was greater than that of perception users at 8 (P<C.01) and 10 (P<.01) yr. of age.
Second trial‑Fig. 3 shows main parts of the results in the second probe trial. The
Utilization of Perceptual and Conceptual Attributes in Children s Discrimination Learning 215
6 yr. 8 yr. 10 yr.
Fig. 3 Performances in the second probe trial
same statistical analyses as those for the first probe trial were applied and very similar results were obtained. The only exception was that for the similar‑category task the difference in percentages between perception and concept users was not significant at 8 yr. of age.
To see the consistency in the use of attributes, the attributes used in the 丘rst and the second probe trials were examined for each subject. The numbers of the subjects who changed the attributes used in the hrst and the second probe trials were 10 (10.4^), 15 (15.6%), and 8 (8.3%) at 6, 8, and 10 yr. of age, respectively. The numbers of the subjects who changed from perception to concept users were 4 (5.6^) and 9 (¥2. b%) for
the similar‑ and dissimilar‑category tasks, respectively, and those who changed from conceptto perception users were ll (15.3%) and 9 (¥2.b%) in the same order. As a whole, only 33 (¥¥. 4%) of 288 subjects changed the attributes used in the distrimination task in the course of learning.
Verbalization‑Verbalization data after the second probe trial were analyzed for the subject who used consistently either perceptual or conceptual attribute in the two probe trials. Although various verbalizations were obtained, the followings were designated as correct verbalizations; "Round", "Slender", and "Long" for the perception users, and
"Fruit", "Vegetable", "Something to eat", "Musical instrument", and "Something to sound"
for the concept users.
A2 (Task)く2 (Attribute)く3 (Age) analysis of variance for the percentages of the
subjects who verbalized correctly their responses was performed by the use of arcsin trans‑
formed scores. The results showed that the main effect of attribute (^2‑16. 72, df‑l), the
age by attribute interaction (% ‑ 18. 61, df‑2),and the task by attribute interaction (^2‑10. 06,
dj‑¥) were significant at ¥% level. The percentage of correct verbalizers was greater for
the concept (63.2%) than for the perteption (35.¥%) users. As is shown in Fig. 4, the
216
Takeshi SUGIMURA6 yr. 8 yr. 10 yr.
Fig. 4 Age by attribute interaction (Percentages of correct verbalizers)
significant age by attribute interaction showed that the percentages of correct verbalizers increased with age for the concept users whereas those for the perception users remained constant through these ages. The signi丘cant task by attribute interaction showed that for the dissimilar‑category task the percentage of correct verbalizers was significantly greater for the concept (70.ァ%) than for the perception users (25. 8%) while there was no significant
difference between the concept (55. ¥%) and the perception users (41. ¥%).
Discussion
The hndings in the probe trials were that (a) the probability to use perceptual attributes decreased and that to use conceptual attributes increased with age and that (b) the conceptual attributes were more often used in the dissimilar‑category task than in the similar‑category task. These support the present hypothesis and are in line with the results obtained in the sorting task (Sugimura, 1979).
According to Bruner et al. (1966) and Gollin and Saravo (1971), it was expected that a developmental shift in the use of attributes would occur from the stage dominated by the use of perceptual attributes to the stage dominated by the use of conceptual attributes.
Although the present results showed that the use of perceptual attributes decreased evidently with age, it was not dominant even at 6 yr. of age. In fact, the percentage 'of perception users was about equal to that of concept users at 6 yr. of age. Therefore, the developmental shift mentioned above was not con后rmed in the present study. If children below 6 yr. of age would serve as the subjects, the shift might be obtained. The present results suggest that there is the transition around 6 yr. of age.
It was confirmed that the category similarity of instances affected the use of attributes
Utilization of Perceptual and Conceptual Attributes in Children s Discrimination Learning 217
in learning the discrimination task. In relation to this, Sugimura (1977) assessed children's preferences for perceptual and conceptual attributes in a matching‑to‑sample task, and found that children showed more conceptual preference in the dissimilar‑category task than in the similar‑category task. It is generally assumed that the subjects more easily discriminate and attend to the conceptual attributes of instances when they are given the dissimilar‑
category task than when they are given the similar‑category task. The present study suggests that the use of attributes in the discrimination learning depends not only on children's cognitive develoment but also on the nature of the tasks.
The numbers of the subjects who correctly verbalized their choice responses were greater for the concept than for the perception users. This may be explained by assuming that the concept users selected each instance through covert verbalizations such as fruit, vegetable or food, whereas the perception users responded directly to each instance at the perceptual level. The fact that correct verbalzers increased with age only for the concept users can be related to development of conceptual ability. Klausmeier et al. (1974) have proposed four levels of concept attainment, two of which are relevant to the present study.
According to them, in the classificatory level of concepts the individual can generalize at least two different instances as equivalent in some way and can classify a large number of instances as examplers and nonexamplers. In the formal level the individual can give the name of the concept and can accurately designate instances as belonging or not belonging to the group. The present study suggests that 6‑yr.‑old children are at the classihcatory level while 10‑yr.‑old children are at the formal level as to the present materials.
The fact that very similar results were obtained in the two probe trials con丘rms that the attributes used in the distrimination task was fairly consistent through learning stages. The feedback information on Trial 1 also had no differential effect on the probe performances.
This suggests that traditional procedure where correct and incorrect instances are predeter‑
mined by the experimenter will produce similar results to the ones obtained in the present study.
References
Bruner, J.S., Olver, R.R., & Green field, P.M., et al. 1966 Studie∫ in cognitive gγowth. John Wiley &
Sons.
Gollin, E.S., & Saravo, A. 1971 A developmental analysis of learning. In J. Hellmuth (Ed.)j Cognitive studies. Vol. 1. New York: Bruner/Mazel, pp. 57‑90.
Klausmeier, H.J., Ghatala, E.S., & Frayer, D.A. 1974 Conceptual learning and development: A cognitive view. New York : Academic Press.
Sugimura, T. 1977 Conceptual preferences among preschool children as a function of instruction and category similarity. Bulletin of Nara Univer∫ity of Education, 26, 1, 141‑147.
Sugimura, T. 1979 Utilization of perceptual and conceptual attributes in children's sorting behavior as a function of age and category similarity. Psychological Reports, 45, 171‑174.
Sugimura, T., & Kunimoto, S. 1976 Effects of perceptible attributes of instance on learning conceptual
218
Takeshi SUGIMURAand half‑conceptual sorting tasks in children. Bulletin of Nara Univertity of Education, 25, 1, 137‑lヰs.
Sugimura, T., & Terao, Y. 1976 Concept utilization in children's sorting behavior as a function of age.
Psychological Reports, 39, 1039‑1042‑