Thalamus is a common locus of reading, arithmetic, and IQ: Analysis of local intrinsic functional properties
Introduction
Reading and arithmetic are basic achievement skills that influence individuals’ success at school and beyond. As these domain-specific achievement skills are correlated with general intellectual abilities (indexed by IQ scores) (Gagné and St Père, 2001, Lambert and Spinath, 2018, Mayes et al., 2009, Peng et al., 2019), neuroimaging studies often control for the effect of IQ (i.e., IQ being entered as a covariate of no-interest or being matched between groups) to identify unique neural underpinnings of the achievement skills and their impairments (Ashkenazi et al., 2012, De Smedt et al., 2011, Eden et al., 2004, Hoeft et al., 2006, Koyama et al., 2011, Pugh et al., 2008, Rosenberg-Lee et al., 2011). Although this analytical practice has been criticized from logical, statistical, and/or methodological perspectives (Dennis et al., 2009), it remains a topic of debate on whether IQ should be controlled for when studying relationships between brain structures/functions and the achievement skills. This lack of consensus in the literature is evident by the fact that majority of most recent neuroimaging studies of reading and arithmetic have still opted to control for IQ (Ashburn et al., 2020, Bulthe et al., 2019, Jolles et al., 2016, Michels et al., 2018, Paz-Alonso et al., 2018).
Some prior studies, addressing the role of IQ in predicting the achievement skills and intervention responses, have indicated that IQ is not a direct cause of either academic achievement (Brankaer et al., 2014, Fletcher et al., 1992, Francis et al., 1996, Murayama et al., 2013) or intervention responses for learning difficulties (Stuebing et al., 2009, Vellutino et al., 2000). Furthermore, neuroimaging studies have demonstrated that activations in core regions involved in the achievement skills (e.g., the left temporoparietal junction for reading) are independent of IQ (Hancock et al., 2016, Simos et al., 2014, Tanaka et al., 2011). These prior findings lead us to think that significant correlations observed between the achievement and IQ tests likely reflect the consequence of both tests measuring common latent factors. Under this circumstance, the use of IQ as a covariate of no-interest could remove some unspecified factors accounting for an achievement skill, and thus potentially producing overcorrected or counterintuitive MRI findings. However, it may be equally misguiding to fail to use IQ as a covariate of interest, which would result in disregarding possible effects of shared factors between achievement and IQ measures on brain activation/connectivity.
Alternatively, both achievement and IQ measures can be simultaneously examined (e.g., an F-test with the two measures of interest) to detect regions where brain signals can be explained by either measure or their combination (Mumford, Poline, & Poldrack, 2015). This approach can answer questions, such as “which regions show significant associations with either measure (e.g., reading or IQ) or both measures”. In particular, the identification of regions common to both measures could help to understand neuromechanisms underlying bidirectional interactions between the achievement and IQ measures. Such bidirectional interactions have been recently appreciated, with mounting evidence from longitudinal studies. Speficially, for reading and IQ relationships, early reading performance predicts later IQ, and early IQ predicts later reading performance (Chu et al., 2016, Ramsden et al., 2013, Ritchie et al., 2015). For arithmetic and IQ relationships, 10-week arithmetic training improves IQ, and 13-week reasoning training improves arithmetic performance (Lowrie et al., 2017, Sanchez-Perez et al., 2017). Most evidently, a meta-analysis of longitudinal studies (Peng et al., 2019) has rendered further evidence that intellectual abilities and achievement skills (both reading and mathematics) predict each other even after controlling for initial performance. Crucially, neural substrates underlying such relations, which can be mediated by shared latent factors in the achievement skills and IQ, cannot be delineated by common analytical practices in the literature, that is, IQ being either covaried out (i.e., controlled for) or excluded from analysis.
In the current resting-state functional MRI (rs-fMRI) study, we address this issue by simultaneously examine both achievement (either reading or arithmetic) and IQ measures. Our primary aim is to explore rs-fMRI correlates common to both the achievement and IQ measures in young adults, whose achievement and IQ scores ranged along a continuum from conventionally impaired to superior performance. We address our aim in a twofold way; (1) entering two covariates of interest – one for an achievement measure (either reading or arithmetic) and the other for Full-Scale IQ (FSIQ) and (2) entering the first principal component (PC1) – reduced from the three measures (reading, arithmetic, and FSIQ) – as the covariate of interest. The first approach uses F-tests, allowing us to detect regions associated with either measure (i.e., specific) or those associated with the common variance explained by the two measures (Mumford et al., 2015). The second approach using principal component analysis (PCA) allows us to explore common regions (Pugh et al., 2013), reflecting the shared variance among the three measures (i.e., two achievement measures and FSIQ), irrespective of the achievement domains.
When analyzing rs-fMRI data, we focus on two data-driven metrics that index local/regional intrinsic functional properties; the first is voxel-wise regional homogeneity (ReHo; Zang, Jiang, Lu, He, & Tian, 2004), and the second is fractional amplitude of low frequency fluctuations (fALFF; Zou et al., 2008). ReHo, which is calculated with Kendall’s coefficient of concordance (KCC), estimates local or short-distance intrinsic functional connectivity (iFC) between the time-series of a given voxel and its nearest neighboring voxels. Jiang et al., 2015, Jiang and Zuo, 2016 have postulated that a higher ReHo value, representing higher synchronization of regional brain activity, indicates higher functional specification in a given region (e.g., the primary visual cortex has the highest ReHo value among regions in the visual ventral pathway). Unlike ReHo, fALFF is a frequency-domain analysis to assess the relative contribution of specific low frequency oscillations to the whole frequency range (Zou et al., 2008). That is, fALFF is a measure of local brain activity, and does not provide any information on functional connectivity. Hence, ReHo and fALFF could be complementary in such that they potentially reveal different brain regions associated with cognitive functions and dysfunctions, although similar results/regions are often reported (Bueno et al., 2019, Hu et al., 2016, Yuan et al., 2013).
Both ReHo and fALFF have successfully detected regions associated with individual differences in cognitive abilities (Kuhn et al., 2014, Yang et al., 2015), clinical diagnoses/traits (Du et al., 2019, Han et al., 2018, Hoexter et al., 2018, Respino et al., 2019, Xu et al., 2015, Xue et al., 2018), and training/experience effects (Koyama et al., 2017, Qiu et al., 2019, Salvia et al., 2019, Wu et al., 2019). However, to date, there are only a handful of studies that have applied these metrics to examination of achievement skills, IQ, and/or their relationships. For reading, Xu et al. (2015) have examined fALFF, with controlling for IQ, and revealed that positive associations between fALFF in reading-related regions (e.g., the posterior superior temporal gyrus) and semantic reading. For arithmetic, Jolles et al. (2016) compared a group of children with mathematical difficulties (i.e., lower arithmetic abilities) and the IQ-matched control group, the first of which was characterized by higher fALFF in the intraparietal sulcus – a core region associated with number processing (Dehaene, Piazza, Pinel, & Cohen, 2003) and arithmetic (Bugden et al., 2012, Dehaene et al., 2004, Jolles et al., 2016, Menon, 2010). Regarding ReHo, no study has explored its whole-brain patterns associated with either achievement or IQ measures (but see Koyama et al., 2017 using a region of interest analysis).
We opt to use data-driven ReHo and fALFF as the primary metrics, rather than seed-based correlation analysis (SCA) that is the most common way to examine resting-state functional connectivity. This is because ReHo and fALFF require no prior knowledge or hypotheses, unlike SCA that requires the selection of seeds (i.e., regions of interest). Investigators typically select seeds based on previous task-evoked fMRI findings in relevant cognitive domains: for example, Koyama et al. (2011) have employed multiple seeds based on meta-analysis studies of reading-related fMRI findings. This selection of seeds is investigator-specific (e.g., seed location, seed size), making SCA vulnerable to bias. In other words, SCA potentially overlooks brain regions that are not selected by investigators, as well as brain regions that are not typically activated during cognitive tasks of interest. For example, when examining auditory processing and its disorders, SCA would typically use seeds located in the primary auditory cortex based on prior task-evoked fMRI results (Bartel-Friedrich et al., 2010, Talavage et al., 2014); however, Pluta et al. (2014) have highlighted that ReHo in the precuneus, a core region of the default mode network (Buckner et al., 2008, Raichle et al., 2001), rather than the auditory cortex, is associated with auditory processing disorders. Based on above-mentioned studies, we hypothesize that the current study, which uses data-driven ReHo/fALFF, could reveal regions outside the networks that had been reported by previous fMRI studies (e.g., task-evoked fMRI, rs-fMRI using SCA) of the achievement skills and IQ. This possibility can be even enhanced given that we simultaneously examine both achievement and IQ measures (e.g., IQ as a covariate of interest in F-tests), rather than controlling for IQ (e.g., IQ as a covariate of non-interest) – the latter is a long-standing common or standard analytic practice in the neuroimaging literature of reading and arithmetic.
Section snippets
Participants
Seventy-two young adults (28 males; mean age 21 ± 1.9 years, age range = 18–25) were selected from a larger study (N = 159) that primarily aimed to examine neural mechanisms of sequence learning and overnight consolidation in adolescents and young adults. The inclusion criteria used here were as follows; (1) young adults between the age of 18 and 30, (2) the completion of a battery of standardized tests measuring cognitive abilities, at least word reading, arithmetic, and IQ, (3) the completion
ReHo results from the first rs-fMRI
Table 2 and Fig. 2 summarize significant ReHo results from the three models. Both F-tests (i.e., the first and second models) highlighted the thalamus with the peak voxels located in the left pulvinar (the “LW-Thalamus” cluster for the F-test with LW & FSIQ; the “Calc-Thalamus” cluster for the F-test with Calc & FSIQ). Additionally, the PC1 scores were negatively associated with the thalamus (the “PC1-Thalamus” cluster).
ReHo-behavior relationships
We plotted the mean ReHo values (i.e., the first rs-fMRI data) extracted
Discussion
The current study simultaneously examined the achievement (reading, arithmetic) and IQ measures in young adults, aiming to identify MRI correlates of their common factors. For this aim, we used F-tests, into each of which an achievement measure and FSIQ were entered, as well as investigating the effect of PC1 among the three measures (reading, arithmetic, and FSIQ). The main finding, which was reliable across these analytic models, is that lower ReHo in the thalamus (the peak voxel in the left
Limitations
There are several limitations in the current study. The most evident was a lack of task-evoked fMRI data in the domain of reading, arithmetic, and IQ/reasoning, which restricted direct comparisons of ReHo-behavior relationships during rest and task. Our primary ReHo finding sits in the thalamus, which is the region exhibiting the most dynamic differences in functional network configuration between rest and task (i.e., more globally connected during task than at rest). Thus, examination of both
Conclusions
We simultaneously examine both achievement (reading, arithmetic) and IQ measures, using rs-fMRI metrics that characterize local intrinsic functional properties. The main finding highlights that ReHo (i.e., local functional connectivity) in the thalamus, particularly the left pulvinar implied in selective attention, is a common neural correlate or convergence site for cognitive variation in reading, arithmetic, and IQ measures. Specifically, the higher the ReHo, the lower the performance on all
Authors’ contributions
MSK completed the data analysis, designed the figures, and drafted the original manuscript. MSK and KRP interpreted the results. KRP supervised the study. PJM and WEM designed the MRI protocol, supervised the MRI data collection, and monitored the MRI data quality. MPM supervised the MRI data preprocessing and MRI data analysis. All authors contributed to the final manuscript.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This work was primarily supported by NICHD R01HD065794 to KRP. It was supported in part by gifts to the Child Mind Institute from Phyllis Green Randolph Cowen (MPM is the Phyllis Green and Randolph Cowen Scholar) and NIMH U01MH099059 to MPM. It was also supported in part by the NIMH Intramural Research Program (PJM; 1ZIAMH00278). We thank Airey Lau, Alexis Lee, Bonnie Buis, Morgan Bontrager, Jocelyn Springfield, and Annie Stutzman for behavioral assessment, recruitment, data collection, and
References (199)
- et al.
Is 2+2=4? Meta-analyses of brain areas needed for numbers and calculations
Neuroimage
(2011) - et al.
Visuo-spatial working memory is an important source of domain-general vulnerability in the development of arithmetic cognition
Neuropsychologia
(2013) - et al.
A reproducible evaluation of ANTs similarity metric performance in brain image registration
Neuroimage
(2011) - et al.
A component based noise correction method (CompCor) for BOLD and perfusion based fMRI
Neuroimage
(2007) - et al.
Numerical magnitude processing deficits in children with mathematical difficulties are independent of intelligence
Research in Developmental Disabilities
(2014) - et al.
The role of the left intraparietal sulcus in the relationship between symbolic number processing and children's arithmetic competence
Developmental Cognitive Neuroscience
(2012) - et al.
Multi-method brain imaging reveals impaired representations of number as well as altered connectivity in adults with dyscalculia
Neuroimage
(2019) - et al.
Left inferior parietal lobe engagement in social cognition and language
Neuroscience & Biobehavioral Reviews
(2016) - et al.
Microarchitecture and connectivity of the parietal lobe
Handb Clin Neurol
(2018) - et al.
The creation of a brain atlas for image guided neurosurgery using serial histological data
Neuroimage
(2006)
Intrinsic and task-evoked network architectures of the human brain
Neuron
Thalamic mechanisms in language: A reconsideration based on recent findings and concepts
Brain and Language
Effects of problem size and arithmetic operation on brain activation during calculation in children with varying levels of arithmetical fluency
Neuroimage
Arithmetic and the brain
Current Opinion in Neurobiology
A dual-networks architecture of top-down control
Trends in Cognitive Sciences
A core system for the implementation of task sets
Neuron
Neural changes following remediation in adult developmental dyslexia
Neuron
Auditory skills and brain morphology predict individual differences in adaptation to degraded speech
Neuropsychologia
When IQ is controlled, does motivation still predict achievement?
Intelligence
Local temporal variability reflects functional integration in the human brain
Neuroimage
How common are WM deficits in children with difficulties in reading and mathematics?
Journal of Applied Research in Memory and Cognition
Functional neuroanatomy of three-term relational reasoning
Neuropsychologia
Accurate and robust brain image alignment using boundary-based registration
Neuroimage
Shared temporoparietal dysfunction in dyslexia and typical readers with discrepantly high IQ
Trends in Neuroscience and Education
Low frequency fluctuation of brain spontaneous activity and obsessive-compulsive symptoms in a large school-age sample
Journal of Psychiatric Research
The inferior parietal lobule and temporoparietal junction: A network perspective
Neuropsychologia
Math practice and its influence on math skills and executive functions in adolescents with mild to borderline intellectual disability
Research in Developmental Disabilities
Improved optimization for the robust and accurate linear registration and motion correction of brain images
Neuroimage
Pulvinar contributions to the dorsal and ventral streams of visual processing in primates
Brain Research Reviews
RESCALE: Voxel-specific task-fMRI scaling using resting state fluctuation amplitude
Neuroimage
Competition between functional brain networks mediates behavioral variability
Neuroimage
Selective attention without a neocortex
Cortex
Individual differences in adult reading are associated with left temporo-parietal to dorsal striatal functional connectivity
Cerebral Cortex
Adults with intellectual disability and attention deficit hyperactivity disorder: Clinical characteristics and medication profiles
British Journal of Learning Disabilities
Working memory and executive function profiles of individuals with borderline intellectual functioning
Journal of Intellectual Disability Research
Cerebellar function in children with and without dyslexia during single word processing
Human Brain Mapping
Weak task-related modulation and stimulus representations during arithmetic problem solving in children with developmental dyscalculia
Developmental Cognitive Neuroscience
Abnormal development of sensory-motor, visual temporal and parahippocampal cortex in children with learning disabilities and borderline intellectual functioning
Frontiers in Human Neuroscience
The effect of preterm birth on thalamic and cortical development
Cerebral Cortex
Development of cortical microstructure in the preterm human brain
Proceedings of the National Academy of Sciences of the United States of America
Human pulvinar functional organization and connectivity
Human Brain Mapping
Development of fMRI tests for children with central auditory processing disorders
Vivo
The development of selective attention in children with attention deficit hyperactivity disorder
Journal of Abnormal Child Psychology
The brain's default network: Anatomy, function, and relevance to disease
Annals of the New York Academy of Sciences
Regional dynamics of the resting brain in amyotrophic lateral sclerosis using fractional amplitude of low-frequency fluctuations and regional homogeneity analyses
Brain Connectivity
Predicting children's reading and mathematics achievement from early quantitative knowledge and domain-general cognitive abilities
Frontiers in Psychology
A developmental fMRI study of reading and repetition reveals changes in phonological and visual mechanisms over age
Cerebral Cortex
The segregation and integration of distinct brain networks and their relationship to cognition
Journal of Neuroscience
Novice readers: The role of focused, selective, distributed and alternating attention at the first year of the academic curriculum
Iperception
Cited by (14)
The mathematical brain at rest
2023, Current Opinion in Behavioral SciencesAtypical brain structure mediates reduced IQ in young adults born preterm with very low birth weight
2023, NeuroImageCitation Excerpt :The thalamus acts as a relay station for almost all incoming sensory information to the cerebral cortex, and has a key role in perception, learning (Jaepel et al., 2017), and consciousness (Lagercrantz and Changeux, 2009). A recent resting-state functional MRI study identified the thalamus as common locus for IQ, as well as domain-specific cognitive skills such as reading and arithmetic (Koyama et al., 2020). The time between very preterm birth and term-equivalent age is a critical period for the establishment of functional thalamocortical connections.
The Intelligence Quotient-math achievement link: evidence from behavioral and biological research
2022, Current Opinion in Behavioral SciencesCitation Excerpt :This may underestimate the possible effects of common factors between mathematical skills and IQ. In order to identify the common neural correlates of mathematical skills and IQ, a recent resting-state fMRI study examined mathematical achievement (arithmetic) and IQ measures simultaneously in 72 young adults [43••]. Local intrinsic functional connectivity analyses indexed by regional homogeneity (ReHo) showed that lower ReHo in the pulvinar nucleus was associated with higher performance on the mathematical and IQ measures.