Lack of Coherence and Divergent Thinking: Two Sides of the Same Coin in Artistic Talent? NICOLA RYDER, LINDA PRING, and BEATE HERMELIN Goldsmiths College, University o f L o n d o n
The aim of this study was to identify some of the underlying cognitive components of artistic talent. The participants in this study were an artistically gifted group of young adults, and a group of controls with no such talent but matched on intelligence. Two components were investigated, one was a preference for local over global processing and others were measures of divergent thinking. The results supported the hypothesis that talented artists might, in contrast to others, show a distinctive cognitive style which may be linked to divergent thinking.
INTRODUCTION ome of the recent research into cognitive components of specific talents have been based on the study of the idiots savants phenomenon. The term idiots savant describes those individuals who, in spite of a low level of general cognitive functioning, never the less show outstanding abilities in narrowly defined areas. The savant syndrome is especially prevalent in individuals with a diagnosis of autism (Pring, Hermelin and Heavey, 1995). One domain of talent which has received attention is savant ability for drawing. Furthermore it is generally supposed that artistic talent may function independently of general intelligence (Gardener, 1983) and may be considered a modular ability. It has been suggested that autistic individuals share a cognitive style which reflects "weak central coherence" (Frith, 1989, Filth & Happ6, 1996). Weak central coherence is characterised by a tendency to draw away from the integration of information in order to obtain an overall contextual meaning and a coherent representation. This contrasts with the more normal cognitive style which functions to provide meaning to events. Examples of such functioning include extracting the overall gist of a story rather than the exact words, or seeing a picture in terms of an overall gestalt rather than its constituent parts. Filth's prediction concerning the weak central coherence in autism has been supported by a number of studies, Filth and Snowling (1983) found autistic children displayed a significant disadvantage in the context dependent processing of ambiguous homophones, Happ6 and Frith (1996) showed autistic individuals performed signifi-
S
Current Psychology: Developmental 9 Learning ~ Personality ~ Social Summer 2002, Vol. 21, No. 2,
168-175.
Ryder, Pring , and H ermelin
169
cantly better than matched controls on their ability to identify hidden figures from a complicated pattern. Importantly Frith's theory also explains why autistic individuals usually do well on the block design subtest from the Weschler intelligence scale. Shah and Frith (1993) demonstrated that this superior performance was due to an ability to segment the configuration into its constituent parts and thus reconstruct the overall pattem, without the interference of the cohesive nature of the total gestalt. Pring, Hermelin and Heavey (1995) maintain that such a segmented processing used by autistic individuals may result in savant skills found in several domains of talent. Evidence of segmentation is apparent in musically talented savants, who often possess perfect pitch (see Miller, 1989). Pring et al. (1995) argue that this ability to segment an auditory gestalt is similar to the visual segmentation shown on tasks such as the block design and the embedded figures. One study which illustrates the use of segmentation in an artistically gifted savant was conducted by Mottron and Belleville (1993) on the savant artist E.C. Mottron and Belleville identified that when drawing E.C began with secondary detail with "no privileged status of global form," unlike the control who began with the overall construct and added the secondary detail, or parts as the drawing progressed. Mottron and Belleville concluded that E.C appeared to show a lack of hierarchical organisation in perceptual processing, in that rather than an impairment to global processing equivalent status was given to both local and global forms. Pring, Hermelin and Heavey (1995) hypothesised that such segmented processing might not only be the key to the prevalence of autistic individuals amongst savants, but may also link savant specific abilities to modular talents found in the general population. In an investigation of segmentation, Pring et al. found that artistic savants showed comparable performance to normal children who also displayed such talent, whilst both groups performed significantly better than their intelligence and diagnosis matched, non gifted controls. This finding suggests that the ability to process information in local rather than global terms is dominant in individual's with artistic talent independent of a diagnosis of autism. The results of the outlined study suggest that in certain talent domains cognitive mechanisms such as segmentation rather than coherence may be adaptive. This is especially apparent in art where the artist goes beyond the usual perceptual restraints to emphasis different components (Pring et al. 1995). This view is also supported by a finding by Getzels and Csikszentmihalyi (1976), in which artistically gifted students showed superior performance on a hidden shapes task similar to the embedded figures task used by Shah and Frith (1983). Although Getzels and Csikszentmihalyi (1976) were not specifically studying the cognitive components of artistic talent and the talented and non-talented groups were not matched on relevant factors, the results do provide further weight for the view that artistic talent may be, in part, mediated by a segmentation process which gives equal status to both local and global functioning. It can be seen from the above studies that there does appear to be a common perceptual process, independent of intelligence, related to the domain of artistic talent. It would thus be interesting to identify any other areas related to this talent domain
Current Psychology / Summer 2002
170
other than perceptual processing which are talent dependent. One area commonly associated with artistic talent concerns scores on divergent thing tasks (Lloyd-Bostock, 1979) more traditionally thought to measure creativity. High scores on such tasks are largely dependent on using different types of information out of obvious, confined context and, the formation of novel associations (Guildford, 1956). A common example of a divergent thinking test is the alternative uses tests, whereby as many alternative uses have to found for a common everyday object. Such tests involve a multitude of responses which can be quantitatively scored reflecting the four general components of divergent thinking; fluency, flexibility, originality and elaboration (Hocevar, 1981). It may be thought that divergent thinking broadens the contextual domain of the problem space. On the other hand a tendency for segmentation severely restricts this, to such an extent that the wider or any context is lost. Yet there may still be an association between the two processes, for example Happ6 (1994) describes an autistic child who was instructed to name the parts of a toy bed. The child correctly labelled the mattress, bed and quilt but when asked to name the pillow replied "it's a piece of ravioli." Such a lack of contextual cue use on a conceptual level is an example of weak central coherence, however it is also an example of divergent thinking based at the perceptual level. Hence it is possible that divergent thinking is a manifestation of a lack of coherence. Further support for this proposed link between a segmentation strategy, or field independence, as identified by performance on tasks such as the embedded figures test, and divergent thinking, as measured by the remote associates test, have been found in a number of studies. Chada (1985) found that individuals gaining lower scores on creativity tests were significantly more field-dependent than high scorers. This finding is given further weight in a study by Noppe-Lloyd and Gallagher (1977) who again found that field-independent participants were significantly more creative than field-dependent participants. The aim of this study is to identify some of the underlying processes adopted by artistically gifted individuals, focusing on segmentation on one hand and the forming of divergent, context independent associations on the other. For this purpose we will compare gifted art students with intelligence matched non gifted controls, exploring the relationship between segmentation at the perceptual level and divergent thinking. METHOD
Participants Two groups participated in this study, an artistically talented group and a control group with no reported artistic talent. The artistically talented group consisted of 20 fine art students enrolled in a fine art degree at a university with a five star rated department. The mean age was 22 years (SD=2). The control group consisted of 20 psychology students participating in the study in order to gain credits toward a first level course, the mean age was 23 years (SD=1.9). None of the subjects in the control
Ryder, Pring, and Hermelin
171
group had formal art qualifications such as A-levels or G.C.S.E'S. The artistically talented group were matched with the controls on scaled scores obtained from the vocabulary subtest of the Weschler Adult Intelligence scale (1981). The mean score was 12.6 (SD. 1). The scores are commonly used as a measure of verbal intelligence.
Materials The study consisted of five separate tests. Two associated with weak central coherence in visual or perceptual processing, two tests of divergent thinking and a non verbal memory task. The tests used to measure weak central coherence were the Block Design (Weschler, 1981) and the Group Embedded Figures test (Witkin, et al. 1971). The Block Design test consisted of nine identical blocks, measuring 78ram by 78mm. Each block comprised of two sets of three different patterns. Subjects were required to construct a series of eight patterns, represented by two dimensional pictures, using the blocks. The Group embedded figures test comprised of three sections. One trial section and two test conditions with nine figures in each. The aim of this task was to identify a simple figure among a larger complex pattern. In both tests superior performance is dependant on the segmentation of the larger picture, thus the use of local over global processing. The two divergent thinking tasks were adapted from Getzels and Jackson's (1962) test of creativity. They were obtained from amongst five tests measuring creativity as a cognitive style, unrelated to creativity in the artistic or scientific sense. The first task was a word meaning test in which subjects were asked to give as many definitions or uses for a common word. The three words used were judged by the experimenter to have a number of different definitions as defined by the Oxford Dictionary of English (1995). The words used were; beat, charge and pass. The second task adapted from Getzels and Jackson was concerned with uses of things. The aim of this task was for the subject to identify as many possible uses for a normal everyday object. The objects named were; a brick, a toothpick and a rubber band. The final test used was a visual memory task based on the Benton visual retention task (Anastasi, 1982) to see if there was a difference in the structural memory of the artistically talented group and the non talented group. This test consisted of ten geometrical designs presented on 150mm by 100mm. card, which later had to be reproduced accurately. This was included to establish whether visual memory distinguished between groups.
Procedure Each participant was tested individually, the order of the tests remained the same throughout. Participants were given the divergent thinking tests first, beginning with the word meaning task. They were instructed by the experimenter to give as many definitions or uses for the following word as possible within a two minute time limit. Participants were then given the example of the word "bolt" to fully illustrate the purpose of the task ("bolt as to secure something, a horse bolts, bolt of lightning").
172
Current Psychology / Summer 2002
The answers were scored on the number of different contexts used rather than the total number of answers given, therefore if for the word pass a subject answered pass someone in the street, pass a car and so on, the participant scored only one mark as the answers were in the same context. The results of each word were added together to obtain an overall score. The second test given was the uses for objects. Participants were instructed to give as many uses they could think of for a common everyday object and again had a time limit of two minutes. Again the answers were scored on the different contexts used, hence if a subject gave the uses for a brick as being building a house, building a wall and so on the answers were given only one mark, whereas if the answers "to throw, to build with, to stand on to see over a wall" were given the participant scored three marks. As in the word meaning test participants marks on the three sections were added together to obtain one overall score. The next test given was the memory task. The participant was instructed to remember as many of the designs as possible and reproduce them after they were presented. The designs were presented one after each other for three seconds, after which the subjects were given as much time as necessary to reproduce the designs from memory. Designs were then scored on a three point scale. If the subjects design was identical to the original then three marks were awarded, if one of the main elements of the design was incorrect but the overall design remained two marks were given, if the subject only produced one correct element of the original design one mark was awarded. The block design subtest was then presented to participants, each having to recreate a line drawing using six or nine coloured blocks, within a given time limit. Finally the embedded figures test was presented. This task consisted of a three trials, including a practice trial, whereby the participant was required to locate a simple figure, held in memory, from within a complex pattern. RESULTS
The mean scores and standard deviations for each task are presented in Table 1. The memory scores are out of a total of thirty, as there was a maximum of three points for each of the ten designs. The block design results were scored as instructed in the manual, with a possible maximum of 51. The embedded figure score consists of the total number of hidden shapes correctly identified on the two test sections, with a maximum score of 18. The scores on the divergent thinking tasks derive from the number of answers using different context. The performance of the two groups did not differ on the memory test (t=2,38, .92, n.s). The artistically talented students obtained significantly superior scores than the controls on the block design test (t=6.27, p<.001). This result was replicated in the embedded figures test with the artistically talented group correctly identifying more of the hidden shapes than the controls (t--4.76, p<.001). Finally, for the two measures of divergent thinking, the artistically talented group also produced significantly superior scores on both the word meaning task (t=5.44, p<.001) and the uses of objects (t=5.44, p<.001).
Ryder, Pring, and Hermelin
173
TABLE 1 Mean scores and standard deviations for task performance.
Artists Non-Artists
Memory
Block design
18 SD. 3.73 17 SD. 1.45
45.8 SD.2.3 35.7 SD. 6.6
Embedded figures 16.7 SD.I.5 12.3 SD. 3.9
Word Meaning 14.2 SD. 2.2 9.9 SD. 2.8
Object uses 23.5 SD. 6.8 13.5 SD. 4.6
Within the talent group scores between tests were found not to be correlated while in the non talented group there was only one significant correlation between scores on the embedded figures and scores on the block design (r=-.77, df=18, p<.001). A multiple regression was then carried out on the data to identify whether the performance on the block design test and the embedded figures test predicted overall performance in the combined divergent thinking tests. From such results the regression equation creativity = 6+ (.29 block design) + (.78 embedded figures) was constructed. The analysis of variance produced a significant result for the regression (R=.54, F(2,37)=7.81, p<.005). However the R 2 result of .29 indicates that the predictors only accounted for 29 percent of the overall variance of the combined creativity scores, also the standard error produced was large. Finally the no significant effects were found for either the block design (8=.24, p>.05) or the embedded figures (13=.32, p>.05). One can thus conclude from these results that a number of variables combine to predict creativity and that although tasks based on segmentation account for some of the variance on divergent thinking tasks they alone are not significant predictors of such performance. Finally a discriminant analysis was carried out on the data. The purpose of this analysis was to identify how well scores on the tests predicted the subjects artistic talent. The residual test procedure identified one significant discriminant function, (•2 = 52.3, p<.001). To interpret this discriminant function, the standardised coefficients and the discriminant function variable correlations are used. The correlations are studied to identify the underlying construct the discriminant function represents. The resuits of this analysis identify the scores on the block design as primarily defining the function (r = .57) with meaning (r = .49) and object uses (r=.49) secondarily involved. The standardised coefficients indicate that all three of the variables, block design (coefficient = .72) word meaning (coefficient = .65) and object uses (coefficient = .55) were involved in the analysis, the scores on the embedded figures test were excluded from the stepwise analysis. From these results it appears that the discriminant function, in the categorisation to an artistically talented group, is best characterised by scores on the block design, word meaning and object uses tests. Furthermore the results of this classification were 95% correct in the case of the artistically talented group and 100% correct in classification to the non talented group. The results from this analysis thus show that segmentation and divergent thinking are significant factors in the determination of artistic talent independent of intelligence.
Current Psychology / Summer 2002
174
DISCUSSION In the present investigation of artistic ability or talent we tried to partial out the effects of verbal intelligence by matching groups on this dimension. The results therefore can be regarded as independent of verbal intelligence. With regards to nonverbal intelligence scores, the block design test is a subtest of the Wechsler intelligence test, thus it can be argued that artists do have higher non-verbal intelligence, which may in part, provide an explanation of the results gained on these tests. The very first point that should be made is that artistic talent makes no claim on visual reproduction accuracy as we found no evidence that groups differed in this respect. This result may be primarily because the stimuli were meaningless and thus not only was there no previously existing representation, but also they did not lend themselves to elaboration or imaginable processing which might have in other circumstance have proved relevant. Significant group differences did emerge on the block design test, thus replicating our finding with graphically gifted children and their non gifted controls (Pring, Hermelin and Heavey, 1995). Similarly the artistic students proved better at identifying hidden figures, replicating the finding of Getzels and Csikszentmihalyi (1976). In addition the two tests frequently associated with "creativity" also differentiated between the groups, even though they were verbally based tests that did not require any intrinsic visual arts component. The performance on these two tests together with the scores on the block design test predicted with great accuracy which students were from within the art department. The results of the two segmentation tasks, the block design and the embedded figures tests, support the view that artistic talent is associated with a bias to process not only the unified gestalt, but also its constituent parts. Thus it is easier to both reproduce the target pattern in the block design test and to identify the simple shape in the embedded figures test if the target scene is equally available both in its entirety and in its constituents. As Van Sommers (1994) states, before reality is produced in a drawing, the essential features must first be extracted and reconstructed. It is interesting to speculate that such a style may allow individuals to reconstitute the visual world in entirely new ways in terms of their formal components as for instance as is evident in the works of Cezanne or Picasso (Gombrich, 1986). It is not surprising that artistically talented individuals should score highly on tests associated with creativity. Nevertheless, as mentioned above, these tests were verbal and therefore it can be argued, that the functions tested here are not confined to the domain of talent, but are more global in nature. It appears that the artistically talented individuals were less "captured" by context and less restricted by one particular meaning and it's close associations. Thus weak central coherence may as suggested, operate both in segmentation and divergent thinking tasks. Although the embedded figures task did significantly differentiate between the groups, the multiple regression showed that, in comparison with the block design test, it added no further explanation to the variance. The performance of the artistic group tended to cluster at the top end of the distribution and thus this test showed less sensitivity than the block design test where a wider range of scores pertained.
Ryder, Pring, and Hermelin
175
The results reported here might be regarded as unremarkable since they indicate that artistically gifted individuals have good spatial abilities and are creative. However, that these two functions actually predict artistic ability with a high degree of accuracy was far more surprising. Earlier the point was made that weak central coherence was a crucial contributing style to s a v a n t artistic ability, the present results can be used, not only to support this hypothesis, but also Guildford's original notion that such context independent thinking is measured in his so-called creativity tests. The results thus highlight, perhaps one single cognitive bias that links segmentation and divergent thinking and shows them to be two sides of the same coin, a coin that may be valid in the case of individuals with normal mental capacities including those with talent. With regard to the savant it remains to be established whether they too are not only characterised by a segmentation style but also show commensurate context independent mental processing strategies. NOTE Accepted for publication: 14 April, 2000. Address correspondence to: Department of Psychology, Goldsmiths College, University of London, London SE14 6NW.
REFERENCES Anastasi, A.(1982). Psychological testing. 5th edn. New York: Macmillan. Chada, N. K. (1985). Creativity and cognitive style. Psycho-Lingua, 15, 81-88. Frith, U. (1989a). Autism explaining the enigma. Oxford: Basil Blackwell. pp. 88-102. Frith, U. & Snowling, M. (1983) Reading for meaning and reading for sound in autistic and dyslexic children. Journal of Developmental Psychology, 1,329-342. Gardener, H. (1983) Frames of Mind. Heinemann: London. Getzels, J. W. & Csikszentmihalyi, M. (1976) The creative vision: A longitudinal study of problem finding in art. New York: John Wiley. Getzels, J. W. & Jackson, P. W. (1962). Creativity and Intelligence: Explorations with gifted students. Wiley and Sons: Chichester. Gombrich, E. (1986). The Purpose and Limits of a Psychology of Art. Bulletin of the British Psychological Society, 39, 75-76. Guildford. J. P. (1956). The Structure of Intellect. Psychological Bulletin, 53, 267-293. Happr, F. G. E. (1994). Autism." An introduction to psychological theory. UCL Press: London. Kohs, S. C. (1923). Intelligence Measurement. New York: Macmillan. Lloyd-Bostock, S. M. A. (1979). Convergent-divergent thinking and arts-science orientation. British Journal of Psychology, 70, 155-163. Miller, L. (1989). Musical savants: exceptional skills in the mentally retarded. Hillsdale, New York: Erlbaum. Mottron, L. & Belleville, S. (1993). A study of perceptual analysis in a high-level autistic subject with exceptional graphic abilities. Brain and Cognition, 23,279-309. Noppe-Lloyd, D., & Gallagher, J. M. (1977). A cognitive style approach to creative thought. Journal of Personality Assessment, 14, 85-90. Oxford English Dictionary (1995). Pring, L., Hermelin, B. & Heavey, L. (1995). Savants, segments, art and autism. Journal of Child Psychology and Psychiatry, 36, 1065-1076 Shah, A., & Frith, U. (1983). An islet of ability in autism: a research note. Journal of Child Psychology and Psychiatry, 24, 613-620. Shah, A., & Frith, U. (1993) Why do autistic individuals show superior performance on the block design task? Journal of Child Psychology and Psychiatry, 34, 1352-1364. Witkin, H., Oltman, E., Raskin, E., & Karp, S. (1971). A manual for embedded figures test. Consulting Psychologists press. Wechsler, D. (1981). Wechsler Adult Intelligence Scale. Revised. The psychological co-operation.
