Brain and Language 85 (2003) 427–431

www.elsevier.com/locate/b&l

Dyslexia linked to talent: Global visual-spatial abilityq

arolyi,a Ellen Winner,b Wendy Gray,c and Gordon F. Shermand,*
Catya von K

a
University of Wisconsin-Eau Claire, Wisconsin, USA
b
Boston College and Harvard Project Zero, Chestnut Hill, MA, USA
c
Brown University Medical School, Providence, RI, USA
d
Newgrange School and Educational Outreach Center, Princeton, NJ, USA
Accepted 4 December 2002

Abstract

Dyslexia has long been defined by deficit. Nevertheless, the view that visual-spatial talents accompany dyslexia has grown, due to
reports of individuals with dyslexia who possess visual-spatial strengths, findings of elevated incidence of dyslexia in certain visual-
spatial professions, and the hypothesis that left-hemisphere deficits accompany right-hemisphere strengths. Studies have reported
superior, inferior, and average levels of visual-spatial abilities associated with dyslexia. In two investigations, we found an asso-
ciation between dyslexia and speed of recognition of impossible figures, a global visual-spatial task. This finding suggests that
dyslexia is associated with a particular type of visual-spatial talent—enhanced ability to process visual-spatial information globally
(holistically) rather than locally (part by part).
Ó 2003 Elsevier Science (USA). All rights reserved.

Keywords: Dyslexia; Reading disabilities; Reading disorders; Learning disabilities; Visual-spatial; Impossible figures

1. Introduction domain. In two studies, we observed evidence that dys-

lexia is associated with a specific visual-spatial talent.
The consensus of most dyslexia researchers is that Since the earliest descriptions of dyslexia (Hinshel-
dyslexia is a phonological deficit characterized by diffi- wood, 1909; Morgan, 1896; Orton, 1925), there has been
culty distinguishing speech sounds (phonemes) aurally conjecture that this condition is associated with en-
and as represented in written language (Shaywitz & hanced visual-spatial abilities (Geschwind, 1982; Gor-
Shaywitz, 1999). It would be surprising, therefore, to find don, 1983; Rourke & Finlayson, 1978; Silverman, in
that a condition viewed as a linguistic deficit is also press; Vail, 1990; West, 1991; Winner & Casey, 1993;
characterized by above average ability in a non-linguistic Winner, Casey, DaSilva, & Hayes, 1991). Neurological
evidence is consistent with the possibility that the dys-
q
Funding for this research was provided by a grant from the lexic brain might process visual-spatial information in
International Dyslexia Association awarded to Ellen Winner and an atypical manner (Riccio & Hynd, 1996). Dyslexic
Catya von K
arolyi; additional funding came from a Boston College brains are atypical in the structure, development, and
Faculty Research Expense Grant to Ellen Winner. We thank the organization of their cells (Galaburda, 1993; Galaburda,
following school officials for allowing us to conduct research at their
schools: Larry Brown, Upper School Head, Carroll School, Lincoln,
Sherman, Rosen, Aboitiz, & Geschwind, 1985;
Massachusetts; Judy Nutting, Administrator at the Carroll School; Humphreys, Kaufmann, & Galaburda, 1990; Sherman,
John Dempsey. Principal of the Devotion School in Brookline, 1998; Riccio & Hynd, 1996), as well as at the gross
Massachusetts; Dave OÕHara, Assistant Principal of the Devotion anatomical level (Galaburda & Kemper, 1979; Gesch-
School in Brookline, Massachusetts; and Robert J. Weintraub, wind & Levitsky, 1968; Riccio & Hynd, 1996). Dyslexic
Headmaster of Brookline High School, Brookline, Massachusetts.
And a special thanks goes to Alice Garside. We also thank Victor
brains are also atypical in patterns of hemispheric acti-
Denenberg for his feedback on an earlier draft of this manuscript. vation during both linguistic and nonlinguistic tasks
*
Corresponding author. (Helenius, Salmelin, Service, & Connolly, 1999; Rum-
E-mail address: [email protected] (G.F. Sherman). sey, 1996; Shaywitz & Shaywitz, 1999).

0093-934X/03/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0093-934X(03)00052-X
428 C. von K
arolyi et al. / Brain and Language 85 (2003) 427–431

Investigations into the possibility that individuals hemisphere deficit in dyslexia. Individuals with dyslexia,
with dyslexia are superior in visual-spatial abilities has, should excel at visual-spatial tasks mediated by the
however, yielded conflicting findings. Individuals with right-hemisphere. Global (holistic) processing of visual
reading disorders have been found to have visual-spatial information is typically mediated by the right-hemi-
abilities that are superior (Bannatyne, 1971; Hooper & sphere (Martinez et al., 1997). Therefore, we compared
Willis, 1989; Naidoo, 1972; Rugel, 1974; Sinatra, 1988; adolescents with and without dyslexia on a global vi-
Swanson, 1984), inferior (Bannatyne, 1971; Benton, sual-spatial task requiring holistic inspection (von
1984; Johnston & Ellis Weismer, 1983; Morris et al., K
arolyi, 2001).
1998; Naidoo, 1972; Rourke, 1985), and comparable Participants looked at line drawings of figures and
(Koenig, Kosslyn, & Wolff, 1991; Rudel & Denckla, indicated which represented possible objects (that could
1976; Rugel, 1974; Siegel & Ryan, 1989; Sinatra, 1988; exist in three-dimensional space) and which represented
Smith, Coleman, Dokecki, & Davis, 1977) to controls. impossible ones. Impossible figures represent objects
One source of this inconsistency likely comes from that seem to be three-dimensional, but could not actu-
the variety of measures used to assess visual spatial ally exist in three-dimensional space. Images of impos-
abilities. For example, Swanson (1984) found disabled sible figures can be seen in the art of M.C. Escher, whose
readers were better than skilled readers at reproducing work, ‘‘Ascending and Descending,’’ for example, de-
unnamed complex visual forms. Rudel and Denckla picts an impossible staircase. Scanning an impossible
(1976) employed a pattern matching tasks involving figure part by part, but failing to integrate the parts,
stimuli made up of two to seven small dots i.e., [ * * * * * ] yields the misperception that the figure is possible
compared with [ * * * * * ], and found no group differ- (Mottron & Belleville, 1993). Only by scanning globally
ences. Siegel and RyanÕs (1989) study revealed poorer does one recognize that the parts conflict and that the
performance on Block Design, a subtest of the WISC-R, figure is therefore impossible (Schacter, 1992). The im-
by those with reading disabilities. It is doubtful that possible figure illusion is created by lines with deceptive
these three tasks all tap the same visual-spatial processes. connections resulting in ambiguities, such as whether a
Our own research has yielded conflicting results on set of lines represents space or a plane. The impossible
this question. In an effort to differentiate among visual trident in Fig. 1 provides an example. If one attends only
spatial processes we administered a wide range of spatial to the lower part of the figure, the central two lines are
tasks to individuals with and without dyslexia. We se- seen as the edges of a cylinder; if one attends only to the
lected tasks that would allow us to compare group upper part of the figure, they are seen as the edges of an
performance on spatial orientation, spatial visualiza- empty space between two structures. To recognize that
tion, figural flexibility, closure speed, visual search, the figure is impossible, one must inspect the figure
spatial reference memory, perceptual organization, and holistically, integrating both the upper and lower parts
visual memory. We also included a set of tasks we cre- to discern the structural ambiguity of the two central
ated in order to tap ‘‘real world’’ visual spatial ability lines.
(i.e., drawing oneÕs hand and solving a three-dimen- The findings of this study were surprising: individuals
sional puzzle) as well as a set of spatial word problems. with dyslexia proved significantly faster at recognizing
Contrary to prediction, we found that the dyslexia impossible figures as impossible, and their greater speed
group was for the most part either inferior or equivalent
to the control group (Winner et al., 2001). Specifically,
on the spatial reference memory tasks and spatial word
problems, the dyslexia group performed worse than
controls. On the spatial orientation, spatial visualiza-
tion, figural flexibility tasks, we obtained mixed results
with the dyslexia group performing worse or the same as
controls. On the closure speed and visual search tasks,
the dyslexia group performed the same as controls. On
the configural style component of the Rey-Osterrieth,
fewer females in the dyslexia group than those in the
control group drew the recalled figure using a global
configural style, but the males in the dyslexia group
performed the same as did the males in the control
group.
Fig. 1. The impossible trident illusion. If one attends only to the lower
Our failure to demonstrate a superiority in visual-
part of the figure, the central two lines are seen as solid; if one attends
spatial processing associated with dyslexia might have only to the upper part of the figure, they are seen as the edges of an
been due to not having administered the right kind of empty space. In the context of the entire figure, the two central lines
visual-spatial task. Linguistic problems suggest a left- generate structural ambiguity.
 C. von K
arolyi et al. / Brain and Language 85 (2003) 427–431 429

did not lead to a sacrifice in accuracy (von K
arolyi, key selection. Participants practiced the task with 14
2001). This finding suggests that the deficit of dyslexia is figures (seven impossible and seven possible figures)
accompanied by a very specific kind of visual-spatial presented in a fixed random order. Instructions were
talent: rapid and accurate holistic inspection. The pres- reviewed, and the task begun. Figures were presented
ent study was undertaken to determine whether this twice in an order randomized for each participant.
finding could be replicated. Figures remained on the screen for a minimum of
250 milliseconds (ms). As soon as a response key was
pressed, the screen went blank for 250 ms. Altogether,
2. Methods the recitation of the instructions, the practice session,
and task itself took approximately 10 min.
2.1. Participants To reduce violation of the heterogeneity assumption
of least squares analytic methods such as ANOVA, Log
Middle and high school students with dyslexia 10 transforms were used to diminish the positive skew
(n ¼ 29, 17 male and 12 females) from an independent characteristic of response time data (Kirk, 1968/1995).
school for students with dyslexia were compared to Response time data were then standardized to facilitate
public school students (n ¼ 35, 18 males and 17 females) the identification of outliers. Analyses were performed
for global processing of visual-spatial information. on the resultant z-scores. Because sex differences are
Participants were primarily middle class and Caucasian. often found on visual spatial tasks (Linn & Petersen,
The dyslexia and control groups were distributed rea- 1985), sex differences were analyzed and some found,
sonably equally across grade level and age. The control but these did not differ by group membership. The re-
group attended public school and, according to self-re- sults of the sex analyses, therefore, are not reported
port, had no prior diagnosis of any learning disabilities. here.
Participants were paid a modest sum for their partici-
pation.
3. Results
2.2. Materials and procedures
A two-way, Group (dyslexia, control) by Sex analysis
The stimuli consisted of a set of possible (n ¼ 12) and of variance (ANOVA) revealed that individuals with
impossible (n ¼ 11) figures (black line drawings pre- dyslexia were faster than controls at recognizing
sented on a white background) developed by Schacter impossible figures (F ð1; 60Þ ¼ 15:079; MSE ¼ :871; p ¼
and his colleagues (Schacter, Cooper, & Delany, 1990) :0003; d ¼ :862). They were not significantly less
and later matched in complexity by Carrasco and Se- accurate (see Table 1). An examination of the rank or-
amon (1996). Because one figure was removed from the dering of scores supports the finding that individuals in
impossible set when it was determined that it was, in- the dyslexia group performed this task more rapidly.
deed, possible, the match for complexity was slightly Whereas only 45% of the participants had dyslexia, 70%
compromised. The remaining figures were presented in of the top 10 scorers in Study 2 had dyslexia, and 68% of
random order on a computer fitted with color-coded the top 25 scorers had dyslexia. Of the top half of all
keys. A red dot was placed on the option key of the participants, 67% had dyslexia. These striking rank-or-
computer, a green dot on the command key. Partici- dered findings are consistent with the results of the
pants indicated that a figure was impossible (not able to ANOVAs.
exist in three-dimensional space) or possible (able to To rule out the possibility that individuals with
exist in three-dimensional space) by pressing the red or dyslexia were faster than controls because they traded
green key, respectively. To facilitate recall of the asso-
ciation between color and figure type, green was linked
to ‘‘go’’ in the instructions (Go! You could build that) Table 1
Reaction Time and accuracy scores on impossible figures in studies 1
and red with ‘‘stop’’ (Stop! You couldnÕt build that).
and 2
Possible figures served as foils so that a correct response
would not always be to indicate that the figure was of Study 1 Study 2
the impossible type. The experimenter instructed par- n MðSDÞ n MðSDÞ
ticipants to respond as quickly and as accurately as Response Times (ms)*
possible. Key selection and response time (from figure Dyslexia 40 2652 (1528) 29 1736 (766)
presentation until key selection) were recorded by the Control 22 3409 (1550) 35 2945 (2227)
computer. Accuracy
The experimenter read all written materials aloud Dyslexia 40 0.776 (0.172) 29 0.729 (0.175)
(informed consent, debriefing, and instructions) and Control 22 0.829 (0.203) 35 0.790 (0.224)
explained that one could use either or both hands for Note. * Significant group difference.
430 C. von K
arolyi et al. / Brain and Language 85 (2003) 427–431

accuracy for speed, we performed three regressions: one compelling implication of this finding is that dyslexia
within each group and one for both groups combined. should not be characterized only by deficit, but also by
These analyses assessed the extent to which a speed-ac- talent. Global visual-spatial processing (what we refer to
curacy trade-off was responsible for the results. Such a as ‘‘holistic inspection’’) may underlie important real-
trade-off would be indicated by a positive linear rela- world activities such as mechanical skill, carpentry, in-
tionship between total response time and number of vention, visual artistry, surgery, and interpreting X-rays
correct responses (i.e., longer response times associated or magnetic resonance images (MRI).
with greater accuracy, shorter response times with lower Linking dyslexia to talent casts this condition in far
accuracy). No speed-accuracy trade-off was revealed more optimistic light than linking it to a deficit only.
for the dyslexia (r ¼ 0:264; p ¼ 0:158) or control Further research in our lab is now being conducted to
(r ¼ 0:094; p ¼ 0:597) groups; nor was such a trade- investigate whether individuals with dyslexia are faster
off found for the participant pool as a whole on other global visual-spatial processing tasks besides
(r ¼ 0:063; p ¼ 0:621). To control for ceiling effects identification of impossible figures, and whether they
(restriction of the upper end of the range of possible possess other kinds of talent beyond rapid global visual-
scores that could conceal a linear trend), we removed all spatial processing. The discovery of talent associated
participants with perfect accuracy scores and repeated with dyslexia may eventually lead to more effective ed-
the regressions. Still no significant correlations or linear ucational strategies and help guide individuals with
relationships were revealed (r ¼ 0:081; p ¼ 0:551). dyslexia to professions in which they can excel.
Thus the dyslexia groupÕs faster identification of im-
possible figures was not at the expense of accuracy.
References

4. Discussion Bannatyne, A. (1971). Language, reading and learning disabilities:

Psychology, neuropsychology, diagnosis, and remediation. Spring-
field, IL: Charles C Thomas.
These results suggest that individuals with dyslexia Benton, A. L. (1984). Dyslexia and spatial thinking. Annals of
have superior global visual-spatial processing ability. Dyslexia, 34, 69–85.
This conclusion is based on their recognizing impossible Carrasco, M., & Seamon, J. G. (1996). Priming impossible figures in
figures more rapidly, but no less accurately than those the object decision test: The critical importance of perceived
stimulus complexity. Psychonomic Bulletin and Review, 3(3), 334–
without dyslexia. 351.
Many components make up visual spatial skill and Galaburda, A. M. (1993). Dyslexia and Development: Neurobiological
levels of ability within individuals can vary widely across Aspects of Extra-Ordinary Brains, (Ed.). Cambridge, MA: Harvard
areas. For example, one can have a wonderful sense of University Press.
design and a poor sense of color, or excellent attention Galaburda, A. M., & Kemper, T. L. (1979). Cytoarchitectonic
organization of the human auditory cortex. Journal of Comparative
to detail combined with a poor sense of proportion. Neurology, 190, 597–610.
Global visual-spatial abilities may be similarly distinct Galaburda, A., Sherman, G., Rosen, G., Aboitiz, F., & Geschwind, N.
from other kinds of visual-spatial abilities. (1985). Developmental dyslexia: Four consecutive cases with
With the exception of the configural style component cortical anomalies. Annals of Neurology, 18, 222–233.
of the Rey-Osterrieth, none of the tasks included in the Geschwind, N. (1982). Why Orton was right. Annals of Dyslexia, 32,
13–30.
Winner et al. (2001) study relied on global visual-spatial Geschwind, N., & Levitsky, W. (1968). Human brain: Left-right
processes. This component of Rey-Osterrieth task is a asymmetries in temporal speech region. Science, 161, 186–187.
recall task which relies on memory for a complex figure. Gordon, H. W. (1983). The learning disabled are cognitively right.
The working memory demands of the Rey-Osterrieth Topics in Learning and Learning Disabilities (April), 29–39.
task make it difficult to sort out visual-spatial from Helenius, P., Salmelin, R., Service, E., & Connolly, J. F. (1999).
Semantic cortical activation in dyslexic readers. Journal of Cogni-
other information processing aspects of the task. In tive Neuroscience, 11(5), 535–550.
contrast, the items in the impossible figures task remain Hinshelwood, J. (1909). Four cases of congenital word-blindness
in the participantÕs visual field, which minimizes demand occurring in the same family. British Medical Journal, 2, 1229–
on working memory. The femalesÕ lower global confi- 1232.
gural orientation on the Rey-Osterrieth task, therefore, Hooper, S. R., & Willis, W. G. (1989). Learning disability subtyping:
Neuropsychological foundations, conceptual models, and issues in
might not be related to their global visual spatial abili- clinical differentiation. New York: Springer-Verlag.
ties, but rather to their working memory. Problems with Humphreys, P., Kaufmann, W., & Galaburda, A. (1990). Develop-
working memory are common among those with dys- mental dyslexia in women: Neuropathological findings in three
lexia (Swanson & Siegel, 2001). patients. Annals of Neurology, 28(6), 727–738.
Given that individuals with dyslexia typically read Johnston, J. R., & Ellis Weismer, S. (1983). Mental rotation abilities in
language-disordered children. Journal of Speech And Hearing
slowly and often exhibit slow naming speeds (Wolf & Research, 26(September), 397–403.
Bowers, 2000), the finding that individuals with dyslexia Kirk, R. E. (1968/1995). Experimental design: Procedures for the
are faster than controls on any task is surprising. The behavioral sciences. Belmont, CA: Brooks/Cole.
 C. von K
arolyi et al. / Brain and Language 85 (2003) 427–431 431

Koenig, O., Kosslyn, S. M., & Wolff, P. (1991). Mental imagery and Joint Conference of the Learning Disabilities Network, Hingham,
dyslexia: A deficit in processing multipart visual objects? Brain and MA.
Language, 41, 381–394. Siegel, L. S., & Ryan, E. B. (1989). Subtypes of developmental
Linn, M. C., & Petersen, A. C. (1985). Emergence and characterization dyslexia: The influence of definitional variables. Reading and
of sex differences in spatial ability: A meta-analysis. Child Devel- Writing: An Interdisciplinary Journal, 2, 257–287.
opment, 56, 1479–1498. Silverman, L. K. (in press). Upside down upside-down brilliance: The
Martinez, A., Moses, P., Frank, L., Buxton, R., Wong, E., & Stiles, J. visual-spatial learner. DeLeon Publishing, Denver, CO.
(1997). Hemispheric asymmetries in global and local processing: Sinatra, R. (1988). Styles of thinking and literacy proficiency for males
Evidence from fMRI. Neuroreport: An International Journal for the disabled in print acquisition. Reading Psychology: An International
Rapid Communication of Research in Neuroscience, 8(7), 1685– Quarterly, 9, 33–50.
1689. Smith, M. D., Coleman, J., Dokecki, P. R., & Davis, E. E. (1977).
Morgan, W. P. (1896). A case of congenital word-blindness. British WISC-R scores of learning disabled children. Journal of Learning
Medical Journal, 2.,1 378. Disabilities, 10(7), 437–443.
Morris, R. D., Stuebing, K. K., Fletcher, J. M., Shaywitz, S. E., Lyon, Swanson, H. L. (1984). Semantic and visual memory codes in learning
G. R., Shankweiler, D. P., Katz, L., Francis, D. J., & Shaywitz, B. disabled readers. Journal of Experimental Child Psychology, 37,
A. (1998). Subtypes of reading disability: Coherent variability 124–140.
around a phonological core. Journal of Educational Psychology, 90, Swanson, H. L., & Siegel, L. (2001). Learning disabilities as a working
1–27. memory deficit Issues in education: Contributions from educational
Mottron, L., & Belleville, S. (1993). A study of perceptual analysis in a psychology, 7, 1–48.
high-level autistic subject with exceptional graphic abilities. Brain Vail, P. L. (1990). Gifts, talents, and the dyslexias: Wellsprings,
and Cognition, 23, 279–309. springboards, and finding FolyÕs rocks. Annals of Dyslexia, 40, 3–
Naidoo, S. (1972). Specific dyslexia. New York: Wiley. 17.
Orton, S. T. (1925). ‘‘Word blindness’’ in school children. Archives of von K
arolyi, C. (2001). Dyslexic strength: Rapid discrimination of
Neurology and Psychiatry, 14(November), 581–613. impossible figures. Journal of Learning Disabilities, 34(4), 380–391.
Riccio, C. A., & Hynd, G. W. (1996). Neuroanatomical and West, T. G. (1991). In the mindÕs eye: Visual thinkers, gifted people with
neurophysiological aspects of dyslexia. Topics in Language Disor- learning difficulties, computer images, and the ironies of creativity.
ders, 16(2), 1–13. Buffalo, NY: Prometheus.
Rourke, B. P. (Ed.). (1985). Neuropsychology of learning disabilities: Winner, E., & Casey, M. B. (1993). Cognitive profiles of artists. In G.
Essentials of subtype analysis. New York: Guilford. Cupchik, & J. Lazlo (Eds.), Emerging Visions: Contemporary
Rourke, B. P., & Finlayson, M. A. J. (1978). Neuropsychological Approaches to the Aesthetic Process (pp. 154–170). New York:
significance of variations in patterns of academic performance: Cambridge University Press.
Verbal and visual-spatial abilities. Journal of Abnormal Child Winner, E., Casey, M. B., DaSilva, D., & Hayes, R. (1991). Spatial
Psychology, 6(1), 121–133. abilities and reading deficits in visual art students. Empirical
Rudel, R. G., & Denckla, M. B. (1976). Relationship of IQ and Studies of the Arts, 9(1), 51–63.
reading score to visual, spatial, and temporal matching tasks. Winner, E., von K
arolyi, C., Malinski, D., French, L., Seliger, C.,
Journal of Learning Disabilities, 9(3), 42–51. Ross, E., & Weber, C. (2001). Dyslexia and visual-spatial talents:
Rugel, R. (1974). WISC subtest scores of disabled readers. Journal of Compensation vs deficit model. Brain and Language, 76, 81–110.
Learning Disabilities, 7(1), 46–65. Wolf, M., & Bowers, P. (2000). The question of naming-speed deficits
Rumsey, J. M. (1996). Neuroimaging in developmental dyslexia: A in developmental reading disability: An introduction to the
review and conceptualization. In G. Reid Lyon, & M. J. Rumsey Double-Deficit Hypothesis. Journal of Learning Disabilities, 33,
(Eds.), Neuroimaging: A Window to the Neurological Foundations of 322–324.
Learning and Behavior and Children (pp. 57–77). Baltimore, MD:
Paul H Brookes.
Schacter, D. L. (1992). Understanding implicit memory: A cognitive
neuroscience approach. American Psychologist, 47(4), 559–569.
Schacter, D. L., Cooper, L. A., & Delany, A. (1990). Implicit memory Further reading
for unfamiliar objects depends on access to structural descriptions.
Journal of Experimental Psychology: General, 119, 3–19. Wechsler, D. (1949). Wechsler intelligence scale for children (Manual).
Shaywitz, S., & Shaywitz, B. (1999). Cognitive and neurobiologic New York: The Psychological Corporation.
influences in reading and in dyslexia. Developmental Neuropsychol- Wechsler, D. (1974). Wechsler intelligence scale for children—revised.
ogy, 16(3), 383–384. New York: The Psychological Corporation.
Sherman, G. F. (1998). A section of Symposium: Dyslexia: A Wechsler, D. (1991). Wechsler intelligence scale for children: Third
byproduct of diversity. Paper presented at the Sixteenth Annual edition manual. San Antonio, TX: The Psychological Corporation.