Developmental Changes in California Card Sorting Test Performance

Archives of Clinical Neuropsychology, Vol. 15, No. 3, pp. 243–249, 2000 Copyright © 2000 National Academy of Neuropsychology Printed in the USA. All rights reserved 0887-6177/00 $–see front matter

PII S0887-6177(99)00015-3

Developmental Changes in California Card Sorting Test Performance Kevin W. Greve University of New Orleans

Jeffrey M. Love University of New Orleans

Theodore J. Dickens, Jr. University of New Orleans

Mary C. Williams University of New Orleans

The purpose of the present study was to determine 1) if performance on the California Card Sorting Test (CCST) follows the same developmental gradient as other measures of concept formation and 2) whether the components of concept formation tapped by the CCST are developmentally dissociable. Participants were 68 children and young adults in four age-based groups: 7 to 9 years (n ⫽ 13); 10 to 12 years (n ⫽ 16); 17 to 19 (n ⫽ 20); and, 20 to 22 years (n ⫽ 19). All were of average or higher measured intelligence and screened for neurological, psychiatric, reading and attentional disorder. The findings of the present study suggest that like many other concept formation tasks, CCST performance approximates adult levels by age 10. Further, the different components of concept formation measured by the CCST are dissociable in the youngest children with the development of concept recognition preceding sorting ability. © 2000 National Academy of Neuropsychology. Published by Elsevier Science Ltd

The acquisition of concept formation and categorization abilities in normal children has long been a topic of interest to developmental psychologists. Classic developmental studies using object sorting paradigms suggested that young children do not form categories as do older children and adults (e.g., Inhelder & Piaget, 1964; Vygotsky, 1962). For example, Vygotsky (1934) suggested that “the most important development of thought in adolescence is the change from ‘complex’ types of thinking to conceptual types of thinking” (p. This project was supported in part by a grant to the first author from the University of New Orleans Research Council. The authors would like to thank Nancé Heath for her assistance in data collection. Address correspondence to: K. W. Greve, Department of Psychology, University of New Orleans, New Orleans, LA 70148.

243

244

K. W. Greve et al.

1064). In contrast, Ross (1980) has argued that young children’s categorization abilities are not qualitatively different from adults’ and are actually underestimated by the sorting tasks typically used. Specifically, she presented very young children (ages 24 months or less) with exemplars of sets of categories that represented a continuum of perceptual similarity (i.e., M’s/O’s made of red blocks; animals/men of similar size made of a variety of materials; furniture/food that differed on a number of dimensions including size and material). Half the children were exposed to members of only one category in each set. Later, each familiar stimulus was presented simultaneously with a novel stimulus from the other category. Ross measured the time the subject spent looking at or touching each stimulus and found that, for all categories, the children attended more to the novel stimuli. This result indicated that the children were able to discriminate between categories and that the discrimination was not based solely on perceptual differences; higher order categories, that were based on functional rather than perceptual features, were also recognized. Thus, a methodology employing category recognition demonstrated a higher level of concept formation ability in very young children than have those requiring “productive behavior” like sorting. Ross argued “that conclusions about children’s categorization skills are dependent on the methodology employed, the response required, and the categories presented” (p. 395). Ross’ position that young children’s concept formation abilities may not be qualitatively different from those of older children is somewhat at odds with traditional developmental theory. According to Piaget, development of higher cognitive functions occurs in stages (Crain, 1992). Development of problem-solving begins within the first two years of life with the child’s ability to form cognitive schemes through the use of sensori-motor skills. These schemes are relatively unorganized at this time. The following stage of pre-operations occurs between approximately 2 and 7 years of age. During this time the child’s cognitive ability expands to include the use of symbols. Language also develops at a rapid rate during the pre-operational period, that provides the child with a source of shared symbols to communicate with others. Between the ages of 7 to 11 the concepts of numbers, categories, classes, relations and other aspects utilized in cognitive processes are greatly developed. At this level of cognitive development children learn to process two perspectives simultaneously. In the final stage of formal operations (after the age of 12) cognitive processes are expanded beyond concrete levels of thinking to abstract and hypothetical reasoning. This pattern of cognitive development is roughly tied to neuroanatomical development, particularly patterns of myelogenesis in the frontal lobes (e.g., Diamond, 1990, 1991). Myelogenesis occurs in three general fields of progressive myelination labeled the primordial, intermediate, and terminal fields (Spreen, Risser, & Edgell, 1995). Primordial fields develop before birth and include the myelination of the somesthetic cortex, primary visual cortex, and the primary auditory cortex. The intermediate fields myelinate within the first 3 months after birth and include secondary association areas such as the regions surrounding the primary sensory and motor cortices. During the final stage of myelogenesis the terminal fields develop. These fields are believed to subserve higher cortical functions. More specifically, the myelination process in the prefrontal cortex reaches functional maturity roughly by the age of 12 (Spreen et al., 1995). Studies using neuropsychological measures of concept formation and other higher level “executive” functions (e.g., the Wisconsin Card Sorting Test, Tower of Hanoi) show that adult performance is usually not achieved until about age 10 (e.g., Chelune & Baer, 1986; Welsh, Pennington, & Grossier, 1991) and certainly continues well into the teens (Levin et al., 1991). Thus, improvement in performance on clinical measures of executive function in children follows the gradient described for both cognitive and neuroanatomical development. Nonetheless, it is well-established that several separate, albeit interactive, cognitive operations underlie executive function, problem-solving, and

Concept Formation

245

concept formation skills (Shallice, 1988; Stuss & Benson, 1986) yet most clinical neuropsychological tasks designed to assess these abilities (e.g., Wisconsin Card Sorting Test, Category Test) provide only global measures of performance. As a consequence, those instruments are unable to elucidate the spared and impaired (or in the case of normal children, relatively developed or undeveloped) components of these high level cognitive processes. Most of these clinical measures are unidimensional and all are what Ross would call measures of “productive” behavior. Thus Ross’ data imply that, like the traditional laboratory tests of concept formation used by developmental psychologists, most clinical measures of these abilities may underestimate the higher level abilities of young children. The California Card Sorting Test (CCST; Delis, Squire, Bihrle, & Massman, 1992), was developed specifically to fractionate the various components of concept formation in neuropsychologically impaired adults. By design, the CCST is able to differentiate among concept recognition, concept articulation, and sorting behavior. It may thus be particularly useful in studying the development of concept formation in children. The CCST consists of three sets of stimulus cards. The cards in each set can be validly sorted into groups based on six different sorting rules. Subjects are tested under three different conditions within each set. In Free Sorting (FS), the subject sorts the cards and states the rule upon which each sort was based. In Structured Sorting (SS), the examiner sorts the cards according to each rule and the subject must then verbally describe the rule. In Cued Sorting (CS), the subject sorts the cards based on a cue provided by the examiner. Thus, although still strictly “productive” measures when compared to Ross’s attentional methodology, the additional structure in SS and CS conditions theoretically allows these measures to be more sensitive to the receptive concept knowledge possessed by brain injured persons and children with still developing brains. Two recent studies illustrate the CCST’s ability to fractionate concept formation abilities. Crouch, Greve, and Brooks (1996) presented a right temporal lobectomy patient who was impaired on FS and SS but performed normally on CS. His performance reflected a problem in the generation of or access to concepts (in this case, nonverbal concepts only) not one of simple sorting (measured by CS). In contrast, Greve, Williams & Dickens (1996) found that children with attention-deficit/hyperactivity disorder (ADHD; n ⫽ 24; mean age ⫽ 9.71, SD ⫽ 1.08; mean grade level ⫽ 4.7, SD ⫽ 1.28) had intact concept formation abilities but were impaired relative to non-ADHD children (n ⫽ 39; mean age ⫽ 9.82, SD ⫽ 1.39; mean grade level ⫽ 4.98, SD ⫽ 1.38) on the components of the CCST that required sorting (specifically, the FS and CS conditions). Thus, when the children had only to recognize and articulate the rule by which the cards were sorted (i.e., in SS) they performed normally. On the other hand, even when cues were presented to assist in correctly sorting the cards the ADHD children were impaired. Thus, the CCST is capable of doubly dissociating concept recognition/articulation and sorting abilities. The present study addressed two questions concerning normal development of concept formation as measured by the CCST. First, does performance on the CCST show the same developmental gradient as is seen on other clinical measures of concept formation and executive function? Second, are the different components of concept formation tapped by the CCST dissociable developmentally? In other words, might the CCST be sensitive to the nascent concept formation abilities reported by Ross?

METHODS Subjects Participants in this study were two groups of children and two groups of young adult college students. The groups were comprised as follows: Group 1, 13 children aged 7 to 9

246

K. W. Greve et al.

(mean ⫽ 8.15, SD ⫽ 6.9); Group 2, 16 children aged 10 to 12 (mean ⫽ 10.56, SD ⫽ .81); Group 3, 20 young adults aged 17 to 19 (mean ⫽ 18.3, SD ⫽ .57); and, Group 4, 19 young adults aged 20 to 22 (mean ⫽ 20.79, SD ⫽ .92). The children’s intellectual function was measured using the Kaufman Brief Intelligence Test (Kaufman & Kaufman, 1990) while the adults’ WAIS-R Full Scale IQ’s was estimated from the Shipley Institute of Living Scale (Zachary, 1986; Zachary, Crumptom, & Spiegel, 1985). All subjects were in the average to above average range of measured intelligence (overall mean 107.3, SD ⫽ 8.9, min ⫽ 90, max ⫽ 134) and there were no group differences in measured intelligence (F [3,63] ⫽ 2.20). The children were participants in a summer cognitive development research program. All children had been fully screened for reading disabilities and attentional problems and any children with such problems or history of psychiatric or neurological disease were excluded. The adults were selected from among college students enrolled in introductory psychology courses at a large urban university. Students with a history of reading disability, psychiatric illness, substance abuse, traumatic brain injury or other neurological illness, or whose first language was not English were excluded. Further, they were age-matched to the children plus 10 years. California Card Sorting Test Administration. The CCST consists of three sets of six stimulus cards with a single word printed on each card. The cards in each set can be sorted into two groups of three cards each based on six different sorting rules. The specific rules differ from set to set. Within each set, three rules involve verbal properties of the words (verbal domain) and three involve nonverbal properties of the cards (nonverbal domain rules). In the standard administration, subjects are tested under three different conditions within each set. In the first condition, Free Sorting, the subject sorts the cards and states the rule upon which each sort was based. In the second condition, Structured Sorting, the examiner sorts the cards according to each rule and the subject must verbally describe the rule. In the third condition, Cued Sorting, the subject sorts the cards based on a rule provided by the examiner. This sequence is completed for each of the three sets of stimulus cards. Scores. The CCST generates a large number of potentially valuable scores, however this analysis was limited to the major scores generated in each test condition (five scores). 1) FS-Total Attempts is the number of sorts produced (right or wrong) by each subject (maximum score limited only by time [3 minutes]). 2) FS-Correct Sorts is the number of sorts that matched one of the six correct target sorting rules (maximum 6 points per set). 3 & 4) FS- and SS-Correct Descriptions are the number of verbal descriptions matching the target sorting rule. Verbal descriptions that included both piles of cards (e.g., “these are yellow and these are green”) earned two points while descriptions of only one pile (e.g., “these are yellow and these aren’t”) earned one point (maximum 12 points per set). 5) CS-Correct Sorts is the number of sorts generated in response to the examinerprovided cues. Two points were awarded for correct sorts based on the general rule (e.g., “has to do with the color of the cards”) and 1 point if the specific rule (e.g., “the cards are green or yellow”) was required (maximum 12 points per set).

RESULTS Question 1 (Does performance on the CCST follow the same developmental gradient as other measures of concept formation?) was addressed by examining the change in

Concept Formation

247

scores as a function of group membership. Analysis of variance revealed significant Group effects for all variables (df ⫽ 3, 64; p ⬍ .01) with effect sizes (eta2) exceeding .20 and power exceeding .90 in all cases. See Table 1 for details. Group 4 always performed better than Group 1. Groups 2 and 3 performed better than Group 1 on all scores except FS-total sorts. Group 2 performed more poorly than Groups 3 and 4 on FS-descriptions only. In general, the very young children performed more poorly on the CCST than even the older children. There was a trend for the older children to differ from the young adults on the description scores but not on the sort scores. The second question (Are the components of concept formation tapped by the CCST developmentally dissociable?) was addressed by examining the relationship of FS-correct sorts, SS-correct descriptions, and CS-correct sorts within each age group. These three scores address the essential elements of the CCST: 1) the ability to spontaneously identify a concept and express it via the sorting response (FS-correct sorts); 2) the ability to recognize presorted stimuli and verbally articulate a concept, an inductive reasoning task (SS-correct descriptions); and, 3) the ability to identify the concrete elements of a concept when provided an abstract cue, a deductive reasoning process (CS-correct sorts). Thus, SS and CS serve as control tasks to determine if difficulty in the FS condition was due to problems with either inductive or deductive reasoning processes. To allow for withingroup comparisons between scores, the scores were converted to z-scores based on the norms of Greve, Farrell, Besson, and Crouch (1995). This conversion was done simply to standardize the scores and not necessarily to make statements about impaired function. Careful examination of the means (see Table 2) confirmed the large difference between the young children (Group 1) and the other groups described in the first analysis. The older subjects (Groups 2, 3, and 4) showed no within-group differences on the target variables. Group 1 made significantly fewer correct responses in FS than in SS (t [12] ⫽ ⫺2.81, p ⬍ .01). Finally, the FS and CS conditions were not significantly different.

DISCUSSION The present study sought to determine whether CCST performance follows the same developmental gradient as is seen on other measures of concept and whether the different components of concept formation tapped by the CCST are dissociable developmentally. Consistent with findings for other neuropsychological measures of concept formation and executive function in general, the children 10 years of age and older performed at or near adult level on all aspects of the CCST. All components of concept formation ability measured by the CCST are relatively undeveloped in young school-age children. TABLE 1 Descriptive Statistics and ANOVA Results for CCST Scores

Score

FS-Total Sorts FS-Correct Sorts FS-Correct Descriptions SS-Correct Descriptions CS-Correct Sorts

Group 1

Group 2

Group 3

Group 4

m (SD)

m (SD)

m (SD)

m (SD)

(3.7)a

13.4 10.2 (2.1)a 17.3 (3.7)a 18.1 (4.8)a 30.3 (3.0)a

(3.8)a,b

15.6 13.1 (2.3)b 22.8 (4.7)b 24.0 (5.2)b 33.6 (1.6)b

(4.8)a, b

17.8 14.9 (1.6)b 27.1 (3.7)c 28.6 (4.7)b 34.4 (2.1)b

(4.6)b

19.2 14.8 (1.8)b 27.3 (4.0)c 26.2 (6.2)b 33.5 (2.5)b

F (3,64)

eta2

Power

5.29 18.73 20.63 10.88 8.93

.21 .48 .50 .34 .29

.92 1.00 1.00 .99 .99

means with the same letter are not significantly different at alpha ⬍.01 as indicated by the Sheffé test. means with the same letter are not significantly different at alpha ⬍.01 as indicated by the Sheffé test. cRow means with the same letter are not significantly different at alpha ⬍.01 as indicated by the Sheffé test. aRow

bRow

248

K. W. Greve et al. TABLE 2 Descriptive Statistics and ANOVA Results for CCST Scores

Scores

F-S Correct Sorts SS-Correct Descriptions CS-Correct Sorts

Group 1

Group 2

Group 3

Group 4

m (SD)

m (SD)

m (SD)

m (SD)

⫺2.28 ⫺1.66 (.83)b ⫺1.79 (1.38)a,b (1.15)a

⫺.74 ⫺.63 (.90)a ⫺.29 (.75)a

(1.22)a

(.88)a

.25 .16 (.82)a .09 (.95)a

.19 (.99)a ⫺.25 (1.08)a ⫺.31 (1.14)a

means with the same letter are not significantly different at alpha ⬍.01 as indicated by the Sheffé test. bColumn means with the same letter are not significantly different at alpha ⬍.01 as indicated by the Sheffé test. aColumn

Further, this younger group of children demonstrated a dissociation among the components of concept formation measured by the CCST that was not observed in the other groups. Specifically, they had difficulty spontaneously identifying the concepts represented in the test but improved dramatically when the sorts were executed for them (i.e., in SS). Thus the younger children were relatively able to accurately articulate the rules guiding the examiner’s sorts. They also seemed able to sort stimuli into appropriate groups based on the examiner’s cues but variability in performance obscured differences from FS. Consistent with Ross (1980), these findings suggest that young children have developed some abstract thinking ability that is not captured by simple sorting tasks. It is clear that these abilities do not approach adult levels that are achieved by early adolescence and that additional external structure is required to maximize their concept formation abilities. These findings are consistent with Luria’s ideas concerning development of self-regulation in children (see the discussion in Crain, 1992). Specifically, Luria hypothesized that children first regulate their behavior via the external constraints or the instructions of others and later via self-imposed commands. Within this framework, the SS and CS conditions represent external constraint or instruction, while the FS condition necessitates greater self-regulation. This study demonstrates that the CCST behaves developmentally as do most clinical measures of concept formation and executive function. These changes in high level cognitive ability are consistent with what is known about post-natal brain development (Spreen et al., 1995) and the stages of cognitive development described by Piaget. Unlike most other clinical measures, however, the CCST is able to dissociate among various processes underlying concept formation, a finding that has been demonstrated in both normal and patient samples of adults and children.

REFERENCES Chelune, G. J., & Baer, R. A. (1986). Developmental norms for the Wisconsin Card Sorting Test. Journal of Clinical Experimental Neuropsychology, 8, 219–228. Crain, W. (1992). Theories of Development: Concepts and Applications. Englewood Cliffs: Prentice-Hall, Inc. Crouch, J. A., Greve, K. W., & Brooks, J. (1996). The California Card Sorting Test may dissociate verbal and nonverbal concept formation abilities. British Journal of Clinical Psychology, 35, 431–434. Delis, D. C., Squire, L. R., Bihrle, A., & Massman, P. (1992). Componential analysis of problem-solving ability: Performance of patients with frontal lobe damage and amnestic patients on a new sorting test. Neuropsychologia, 30, 683–697. Diamond, A. (1990). The Development and Neural Bases of Higher Cortical Functions. New York: New York Academy of Sciences. Diamond, A. (1991). Neuropsychological insights into the meaning of object concept development. In S.

Concept Formation

249

Carey & R. Gelman (Eds.), The Epigenesis of Mind: Essays on Biology and Cognition. Hillsdale, NJ: Erlbaum. Greve, K. W., Farrell, J. F., Besson, P. S., & Crouch, J. A. (1995). A psychometric analysis of the California Card Sorting Test. Archives of Clinical Neuropsychology, 10, 265–278. Greve, K. W., Williams, M. C., & Dickens, T. J., Jr. (February, 1996). Concept formation in attention disordered children. Poster presented at the 24th annual meeting of the International Neuropsychological Society, Chicago, IL. Inhelder, B. & Piaget, J. (1964). The early growth of logic in the child. New York: Norton. Kaufman, A. S. & Kaufman, N. L. (1990). Kaufman Brief Intelligence Test. Circle Pines, MN: American Guidance Service. Levin, H. S., Culhane, K. A., Hartmann, J., Evankovich, K., Mattson, A. J., Harward, H., Ringholz, G., Ewing-Cobbs, L., & Fletcher, J. M. (1991). Developmental changes in performance on tests of purported frontal lobe functioning. Developmental Neuropsychology, 7, 377–395. Ross, G. S. (1980). Categorization in 1- to 2-year-olds. Developmental Psychology, 16, 391–396. Shallice, T. (1988). From neuropsychology to mental structure. Cambridge: Cambridge University Press. Spreen, O, Risser, A. H., & Edgell, D. (1995). Developmental neuropsychology. New York: Oxford University Press. Stuss, D. T., & Benson, D. F. (1986). The frontal lobes. New York: Raven Press. Vygotsky, L. S. (1934). Thought in schizophrenia. Archives of Neurology and Psychiatry, 31, 1063–1077. Vygotsky, L. S. (1962). Thought and language. New York: Wiley. Welsh, M. C., Pennington, B. F., & Grossier, D. B. (1991). A normative-developmental study of executive function: a window on prefrontal function in children. Developmental Neuropsychology, 7, 131–149. Zachary, R. A. (1986). Shipley Institute of Living Scale. Revised Manual. Los Angeles: Western Psychological Services. Zachary, R. A., Crumptom, E., & Spiegel, D. E. (1985). Estimating WAIS-R IQ from the Shipley Institute of Living Scale. Journal of Clinical Psychology, 41, 532–540.