A Cluster Analysis of Individual Differences in Reading Skill and Behavioral Lateralization: Associations with Structural Asymmetries Christine 1 Chiarello , Suzanne 1,2 Welcome , & Christiana M. 3 Leonard University of California, Riverside1,University of Western Ontario2, University of Florida, Gainesville3 Three questions were of primary interest: (1) Are there subgroups within the college population characterized by differing profiles of reading and VF/hemisphere asymmetry? The answer can shed light on the relationship between reading skill and degree of lateralization. (2) Do groups identified in the cluster analysis differ in structural asymmetry? If so, this could suggest relationships between behavioral and neuroanatomical profiles that might not be apparent from approaches relying on a priori subject groups. (3) Are there some individuals whose reading behavior and VF asymmetry depart from typically observed outcomes? Do they show unusual neuroanatomical features? If such individuals have less well-regulated trajectories of neural development, we would expect a higher likelihood of unusual neuroanatomical features than for those with more typical profiles. Method PARTICIPANTS: • 100 male, 100 female native English speakers • 18-34 years of age BEHAVIORAL MEASURES: • Word Identification (word reading), Word Attack (nonword reading), and Passage Comprehension subscales from Woodcock Reading Mastery Test - R • Asymmetry scores across the following divided visual field tasks calculated separately for accuracy and reaction time: Lexical Decision, Masked Word Recognition (2 AFC procedure), Word Naming,Nonword Naming, Semantic (manmade vs natural) Decision,Verb and Category Generation BRAIN MEASUREMENTS FROM MRI: • Volumetric MRI scans (1.2 mm thick sagittal slices) on 1.5 GE scanner • Surface area of the planum temporale, planum parietale, Heschl’s gyrus, and the pars triangularis were measured as described in Chiarello, et al (2009). The volumes of the anterior cingulate and the paracingulate sulci were measured as described in Leonard, et al (2009). The planum temporale, Heschl’s gyrus and the paracingulate sulcus are all characteristically larger on the left and some studies have related these asymmetries to linguistic and cognitive skill. 0.5 The following measures, which were not strongly correlated, were used in the cluster analysis (Ward’s method): word attack; accuracy asymmetry for masked word recognition, lexical decision, and verb generation; RT asymmetry for nonword naming, masked word recognition, lexical decision and verb generation. Multivariate outliers (N = 17) were identified by calculating robust Mahalanobis distances. A four-cluster solution (eigenvalue = 0.97) successfully classified the remaining 183 participants, as shown in table below: Poorer readers, somewhat reduced RVF advantages (N = 61) Average readers, large RVF advantages (N = 63) Good readers, reduced RVF advantages (N = 26) Good readers, task-dependent asymmetries (very large RVF advantage for masked word recognition only) (N = 33) All groups showed the expected leftward asymmetries of the planum temporale and Heschl’s gurus, and rightward asymmetry of the planum parietale Clusters Word Attack %ile Word Ident %ile P Comprehension %ile Mean VF Asymmetry (sd) Pl.Temporale Asym2 Pl. Parietale Asym2 Heschl’s Gyrus Asym2 Paracingulate Asym2 Cingulate Asym2 Poorer Reader Average Reader Good Reader Good Reader Low-to-Average VF Asymmetry 31.9 40.7 56.1 -.212 (.206) Large VF Asymmetry 47.3 50.3 63.7 +.249 (.320) Low VF Asymmetry 66.2 57.0 73.7 -.510 (.274) VF Asymmetries Vary by Task 60.8 58.6 73.4 +.163 (.495) .32** .30** -.43** .13** .44** -.26** .40** -.30** .11** .40** -.13* Clus te re d Subje cts 0.4 0.3 0.2 0.1 0 -1.0 to -0.6 -.5 to -0.1 0 to +0.4 +0.5 to +0.9 +1.0 to +1.4 +1.5 to +1.9 Planum Temporale Asymmetry 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 Outlie r s Clus te re d Subje cts -1.5 to -0.1.1 -1.0 to -0.6 -.5 to -.1 0 to +0.4 +.5 to +.9 +1.0 to +1.4 Anterior Cingulate Asymmetry Conclusions .41** -.73** .15** .13 ns -.15 ns • Even among college students, there is diversity in reading ability. We identified 4 reading subgroups with differing VF laterality profiles. Normal leftward perisylvian asymmetries were observed in each of these “typical” profiles. -.37** .15** .40** -.29** • A minority of the sample had reading and VF asymmetry profiles that differed from the rest of the sample. These “outliers” had very good reading comprehension, but demonstrated extreme asymmetry variability in divided visual field reading tasks. 1 Asymmetry expressed as z-score; 0 = average asymmetry; negative = smaller than average asymmetry 2 Coefficient of asymmetry; positive values = leftward asymmetry; * p < .05, ** p < .001 Outliers Compared to Clusters • To determine whether the outliers differed from subjects showing more typical profiles (as captured by cluster analysis), t-tests compared the combined clusters to the outliers (see adjacent table) • The outliers had significantly better passage comprehension, somewhat larger VF asymmetries, and much more variable asymmetries across tasks • The outliers had reduced asymmetry of the planum temporale, with 41% of this group having atypical rightward asymmetry • Outliers also had reduced asymmetry of the anterior cingulate sulcus, with 47% having atypical leftward asymmetry Outlie r s % of group There is substantial individual variation in reading skill, even among college students, but the relationship of this to variation in cortical anatomy and lateralization is unclear. Previous approaches have investigated lateralization differences between groups defined a priori (e.g., sex, handedness), with conflicting results. However, such group differences may not reflect distinct behavioral subtypes that exist within the population of normal readers. Here we identify groups in a bottom-up fashion by patterns in their behavioral data using cluster analysis. Our objective is to identify common variants in the relationship between reading skill and behavioral lateralization, and to examine neuroanatomical correlates. Analyses and Results % of group Objective 0.6 Clustered Subjects Word Attack% 47.3 Word ID% 49.6 P Compreh% 64.3 Mean VF Asym1 -.023 (sd) (.851) Outliers Pl Temporale Pl Parietale Heschl’s G Paracingulate Cingulate .126* -.500 .137 .389 .083* 1 Asymmetry .343 -.417 .129 .291 -.207 55.2 51.2 77.2* +.249* (1.43)** expressed as z-score; 0 = average asymmetry; negative = smaller than average asymmetry *p < .05, ** p < .0001 • The outliers also failed to demonstrate typical leftward planar asymmetry, and typical rightward anterior cingulate sulcus asymmetry, with many showing reversed asymmetries. • We observed an association between atypical structural asymmetry and atypical behavioral profiles. This relationship may reflect a less regulated pattern of neural development, in which random genetic and environmental factors influence cerebral lateralization and behavioral outcomes. • It is notable, however, that this was not associated with poor reading outcomes. References Chiarello, et al. (2009). Neuropsychology, 23, 210-222. Leonard, C.M., et al. (2009). Brain Structure and Function, 213, 553-569. Acknowledgment This research was supported by NIDCD grant 5R01DC6957.