Cerebellum’s role cannot impact reading and Dyslexia


New brain imaging research debunks a controversial theory about dyslexia that can impact how it is sometimes treated, Georgetown University Medical Center neuroscientists say.

The cerebellum, a brain structure traditionally considered to be involved in motor function, has been implicated in the reading disability known as developmental dyslexia.

However, this “cerebellar deficit hypothesis” has always been controversial. The new research shows that the cerebellum is not engaged during reading in typical readers and does not differ in children who have dyslexia.

That finding comes from a new study involving children with and without dyslexia published October 9, 2019, in the journal Human Brain Mapping.

It is well established that dyslexia, a common learning disability, involves a weakness in understanding the mapping of sounds in spoken words to their written counterparts, a process that requires phonological awareness.

It is also well known that this kind of processing relies on brain regions in the left cortex. However, it has been argued by some that the difficulties in phonological processing that lead to impaired reading originate in the cerebellum, a structure outside (and below the back) of the cortex.

“Prior imaging research on reading in dyslexia had not found much support for this theory called the cerebellar deficit hypothesis of dyslexia, but these studies tended to focus on the cortex,” says the study’s first author, Sikoya Ashburn, a Georgetown PhD candidate in neuroscience.

“Therefore, we tackled the question by specifically examining the cerebellum in more detail. We found no signs of cerebellar involvement during reading in skilled readers nor differences in children with reading disability.”

The researchers used functional magnetic resonance imaging to look for brain activation during reading. They also tested for functional connections between the cerebellum and the cortex during reading.

This shows a brain with the cerebellum highlighted

New research shows that the cerebellum (highlighted above) is not engaged during reading in typical readers and does not differ in children who have dyslexia.
The image is adapted from the Georgetown University news release.

“Functional connectivity occurs when two brain regions behave similarly over time; they operate in sync,” says Ashburn.

“However, brain regions in the cortex known to partake in the reading process were not communicating with the cerebellum in children with or without dyslexia while the brain was processing words.”

The results revealed that when reading was not considered in the analysis — that is, when just examining the communications between brain regions at rest — the cerebellum was communicating with the cortex more strongly in the children with dyslexia.

“These differences are consistent with the widely distributed neurobiological alterations that are associated with dyslexia, but not all of them are likely to be causal to the reading difficulties,” Ashburn explains.

“The evidence for the cerebellar deficit theory was never particularly strong, yet people have jumped on the idea and even developed treatment approaches targeting the cerebellum,” says senior author and neuroscientist Guinevere Eden, D. Phil, professor in the Department of Pediatrics at Georgetown University Medical Center and director for its Center for the Study of Learning.

“Standing on a wobble board — one exercise promoted for improving dyslexia that isn’t supported by the evidence — is not going to improve a child’s reading skills.

Such treatments are a waste of money and take away from other treatment approaches that entail structured intervention for reading difficulties, involving the learning of phonologic and orthographic processing.”

In the long run, these researchers believe the findings can be used to refine models of dyslexia and to assist parents of struggling readers to make informed decisions about which treatment programs to pursue.

Funding: This work was supported in part by grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P50 HD040095, R01 HD081078), and the National Center for Advancing Translational Sciences of the National Institutes of Health (TL1 TR001431).

Developmental dyslexia (DD) is traditionally defined as “a disorder in children who, despite conventional classroom experience, fail to attain the language skills of reading, writing and spelling commensurate with their intellectual abilities” (World Federation of Neurology, 1968).

Many attempts have been made to provide fuller or more theory-based definitions of dyslexia, but none has proved as enduring as this initial definition. Very extensive research has taken place over the past three decades, but progress toward a clear understanding, a clear diagnostic system or an effective support system remains elusive.

A critical problem for studying dyslexia is that by the time dyslexia is identified—or even suspected—a child will already be at least 5 years and probably considerably older, and therefore his or her developmental history is lost to detailed investigation.

It is now established that the brain’s primary network structures are developed within the first 2 years of life, both for white matter structural connectivity, and for functional networks including the default mode network, the dorsal attention network and the salience network (Gilmore et al., 2018), and consequently, the early childhood period may prove critical for the understanding of the development of dyslexia.

The need for a developmental analysis of dyslexia is well-established (Karmiloff-Smith, 1998; Goswami, 2003).

There have been two major European longitudinal studies of children born to dyslexic parents (Lyytinen et al., 2004; van der Leij et al., 2013), but these have the inevitable limitations of atypical samples (owing to the need for familial incidence), delay between study design and dyslexia diagnosis (hence the tests undertaken on infants may be outdated by study end) and moreover, it is likely that the parents of participants will be alert to any dyslexia-like issues and may take additional actions. In short, longitudinal studies provide additional converging evidence but cannot in themselves provide the necessary theoretical foundations.

Finally, there is now very extensive evidence that dyslexia overlaps markedly with several other learning disabilities, including Specific Language Impairment (SLI), Attention Deficit and Hyperactivity Disorder (ADHD), and Developmental Coordination Disorder (DCD; Fletcher et al., 1999; Gilger and Kaplan, 2001; Hill, 2001; Boada et al., 2012). Not surprisingly, these issues further complicate the appropriate diagnostic and assessment methods for dyslexia.

Our approach to this series of problems is to take a broader developmental perspective, starting at gestation, and making our way through the developmental processes of the first 5 years of life. We take the view that the extant theories of dyslexia provide a valuable analysis of the reading-level symptoms and that a complete framework should provide an explanation of the developmental processes that lead to this range of symptoms.

The article, therefore, comprises four sections. First, we present an overview of explanatory theories for dyslexia, including our three frameworks for the explanation of dyslexia, namely automatization deficit, cerebellar deficit framework and the subsequent procedural learning framework, with the intention of highlighting potential synergies between the many theories.

We then re-present three experimental studies that are interpretable only within a learning framework, and provide direct evidence of the delay in skill learning exhibited by dyslexic children. The third section attempts to link these findings and theories to the current state of the art in terms of development of language skills and neural networks, highlighting the process of neural commitment considered to underlie much of this developmental trajectory. Finally, we develop the proposal that “Delayed Neural Commitment (DNC),” not just at skill level but also at network level, provides not only a parsimonious characterization of the development of dyslexia but also unique insights into how to mitigate problems caused by this developmental difference.

Levels of Analysis and Theories of Dyslexia

There are many theories for the causes of dyslexia. Accessible overviews of a range of theories were given in Demonet et al. (2004) and also in Nicolson and Fawcett (2008).

A more recent overview focusing on the dominant phonological deficit framework is provided in Peterson and Pennington (2012).

It is beyond the scope of this section to give even a summary of the individual theories, but it is valuable to list some of the more prominent approaches, since it is our intention to try to integrate them within a coherent developmental framework.

When considering theories it is useful to distinguish three levels of explanation (Morton and Frith, 1995): the behavior level (which is directly observable, such as reading), the cognitive level (in terms of underlying theoretical constructs such as memory, language and processing speed) and the brain level (which focuses on neural structures and process).

More recent research suggests the need for two further levels: with the genetic level as the deepest level and the “network” level between the cognitive level and the brain level.

Behavioral Level

Following the standard medical model, the behavioral manifestations may be seen as symptoms of the underlying cause.

The primary symptom of dyslexia is, of course, poor reading. For much dyslexia research, the focus of attention is on reading-related symptoms, and consequently, this research has tended to focus on reading and pre-reading skills. Behavior level theories could, therefore, include lack of opportunity, lack of experience, lack of letter knowledge or lack of “concepts about print” (Clay, 1993).

However, broadening the scope to an attempt to understand the underlying causes brings a range of further potential symptoms into play, in much the same way as in medical diagnosis the symptoms might be fever, but in order to establish the underlying cause a range of further investigations must be made, leading to the establishment of a range of secondary symptoms that, together with the primary symptoms, allow a differential diagnosis of underlying cause.

This is particularly important at the genetic level, where having an appropriate phenotype (symptom) is crucial.

Cognitive Level

Many theories have attempted to explain the behavioral symptom at the next level, namely the cognitive level, thereby providing a potentially causal explanation. The dominant cognitive level theory is the phonological deficit hypothesis (Stanovich, 1988).

The hypothesis proposes that the reading difficulties are attributable to problems in phonological processing, that is, breaking a word down into its constituent sounds. These difficulties cause problems in sound segmentation and also in word blending, both of which are critical for the development of reading and spelling.

There has been extensive research on phonological deficit. However, phonological deficit is by no means the only relevant theory. There are actually many other cognitive level theories, some narrower, some broader. We provide representative examples below. Each one of them has merit—supportive evidence and also successful remediation studies.

The double deficit hypothesis (Wolf and Bowers, 1999) identified two risk factors for reading acquisition: phonological deficit and processing speed deficit.

Children who suffered from a “double deficit” were shown to have a much higher risk of reading problems than children with only one.

Phonological deficit theorists argue that this is best seen as a variant of the phonological deficit hypothesis, and may attempt to subsume phonology, working memory and processing speed within their framework-“deficits in phonological coding [underlie problems in] phonological awareness, alphabetic mapping, phonological decoding, verbal memory, and name encoding and retrieval” (Vellutino et al., 2004, p. 31). A later theory, the phonological access theory (Ramus and Szenkovits, 2008), proposes that the phonological representations are intact but with slower and more effortful access.

The speech rhythm deficit hypothesis (Goswami, 2002) holds that the phonological problems arise from difficulties in perceiving the onset of the amplitude envelope which forms the basis of determining the prosody of an utterance (and hence identifying syllable boundaries).

The visuo-spatial attention deficit hypothesis (Facoetti et al., 2003) attributed reading-related deficits to difficulties in “covert orienting,” that is, preparing to switch attention to a new specific location while still concentrating on the current location.

This is a process required for skilled reading in that the reader is covertly attending to the next words while reading the currently fixated one. A related hypothesis (Bosse et al., 2007) holds that visual attention span is reduced in dyslexia.

Further visual hypotheses relate to fixation accuracy and stability, together with saccadic accuracy. Stein and his colleagues identified eye movement differences (Eden et al., 1994), and several authors have reported disadvantages with visual crowding or advantages for reading with larger fonts (Moores et al., 2011; Schneps et al., 2013).

An independent approach to auditory processing, Tallal et al. (1993) proposed that in common with children with SLI, dyslexic children have specific problems in rapid auditory processing. Both these frameworks have been interpreted at the brain level in terms of the magnocellular deficit hypothesis (see below).

Finally, two hypotheses address the learning processes in dyslexia. A series of studies by Froyen et al. (2011) strongly criticized the phonological deficit account as being a description rather than explanation, and provided evidence that dyslexic children have specific difficulties in integrating the visual letters with their sounds (that is, the visual-auditory cross-modality links are not made automatically).

This visual-auditory integration deficit hypothesis may be seen as a specific instance of the automatization deficit hypothesis (Nicolson and Fawcett, 1990), applied to the reading domain.

The automatisation deficit framework proposes that dyslexic children have difficulties making any skill automatic, whether it is a cognitive skill as in reading, or a motor skill, as in balance or catching. A consequence of the incomplete automaticity is that dyslexic children need to try harder, to “consciously compensate,” even for routine skills that normally-achieving children undertake without effort.

Problems, therefore, become apparent in dual tasks or more complex tasks, where it is not possible to consciously compensate both.

Brain Level

Theories framed at the brain level typically attempt to explain cognitive level deficits in terms of the brain structures that cause them.

The most prevalent brain-level hypothesis for dyslexia is in terms of sensory processing and in particular the “magnocellular deficit” hypothesis.

There is extensive, albeit inconsistent, evidence of specific visual problems relating to detection of low contrast moving visual gratings (Eden et al., 1996), which was attributed to impaired function in the visual magnocellular system.

In an attempt to integrate both visual and auditory magnocellular approaches (Stein, 20012018) has suggested that they may be a pan-sensori-motor abnormality in the magnocellular systems for audition, vision and action.

A broader brain-level theory is our cerebellar deficit hypothesis (Nicolson et al., 19952001).

The cerebellum is a major brain structure, containing over half the brain’s neurons (Brodal, 1981), and with two-way connections to almost all other head and body nervous systems (Bostan et al., 2013).

It has a crystalline structure that supports the development of recurrent circuits (“microcomplexes”) able to scaffold the acquisition and/or execution of a range of motor skills (Ito, 19842008).

The advent of brain imaging highlighted the involvement of the cerebellum in cognitive skills and sensory processing, as well as language through connections to Broca’s area, thereby providing a natural link to the multiple perspectives on dyslexia. Following direct evidence of specific cerebellar deficits in a range of skills (Nicolson et al., 199519992002; Fawcett et al., 1996; Fawcett and Nicolson, 1999) we developed the cerebellar deficit framework for dyslexia, and argued that the framework was able to subsume all the above accounts (automatisation deficit, phonological deficit, speed deficit and sensory integration deficit) at the cognitive level, while providing a potentially causal link to the underlying brain structures and mechanisms.

Of particular interest here, we created the first truly developmental account (see Figure 1) which proposed that a range of factors could be at play in the pre-reading years, and that these could (depending on the number of cerebellar networks involved) lead to a range of symptoms within and beyond reading-related skills. It may be seen from Figure 1 that the major route of impairment is via phonological processing (linked to speech production weaknesses), with additional problems arising from working memory limitations and also (distinctively) from automatisation problems.

These problems give rise (in due course) to the problems of reading and spelling that are the defining features of dyslexia.

The framework has several difficulties, with the major assessment difficulty being that it is extremely difficult to isolate the role of the cerebellum from other brain structures because it works in tandem to optimize performance. Furthermore, given its putative role in the developmental process, standard cross-sectional performance tests lack the necessary investigative power. Third, the cerebellum is a huge structure, and therefore it is critical to identify more specifically which networks are those centrally involved (Stoodley and Stein, 2013).

An external file that holds a picture, illustration, etc.
Object name is fnbeh-13-00112-g0001.jpg
Figure 1
The developmental causal chain (Nicolson et al., 2001).

Fourth, a range of studies have demonstrated that whereas almost all dyslexic children show a phonological deficit, only a subset show difficulties in motor skill, and these children may show additional disorders such as DCD or ADHD (Ramus et al., 2003).

Finally, once the effects of phonological deficits have been accounted for, motor skill deficits do not contribute to the reading deficits (White et al., 2006). We addressed these issues at the time (Nicolson and Fawcett, 20052006) and so should not do so here. Subsequent research has clearly supported the general framework (Alvarez and Fiez, 2018). We discuss this framework and the subsequent Procedural Learning Deficit framework in the following sections.

The Neural Network Level

Recent research in cognitive neuroscience has made it clear that brain regions work together to create skills, and therefore it is important to introduce a level in between the cognitive level and the brain level, namely the network level.

As discussed in section “The Cognitive Neuroscience of Neural Network Development” a biological neural network comprises a group of neurons that are either chemically or functionally related.

Introducing this level addresses some difficulties for brain-level theories. Consider our own cerebellar deficit hypothesis (Nicolson et al., 19952001). It is clear that there is impairment of performance in many skills that involve the cerebellum and indeed there is clear evidence of differences in cerebellar structure (Eckert, 2004; Pernet et al., 2009) and function (Nicolson et al., 1999; Alvarez and Fiez, 2018)

. However, as noted by Zeffiro and Eden (2001), it is quite possible that the cerebellum is actually functioning at normal levels, but that it is really receiving poor quality information from other brain regions such as the senses. It is, therefore, an “innocent bystander” and the true underlying cause lies elsewhere.

This is in fact why we explicitly included other brain networks linked to the cerebellum in our reformulation of the hypothesis (Nicolson et al., 2001). A fuller analysis of the issues involved led directly to our third framework for dyslexia, which is the Procedural Learning Deficit hypothesis (Nicolson and Fawcett, 2007).

The distinction between procedural and declarative systems is long-established in cognitive neuroscience (Squire et al., 1993).

In our research, we had established a range of procedural problems for dyslexia but no declarative problems.

Interestingly, a novel analysis by Ullman (2004) highlighted the fact that there are also procedural and declarative systems for language, with the procedural system corresponding to the “mental grammar.” Ullman and Pierpont (2005) claimed that SLI could be attributed to the abnormal function of the cortico-striatal branch of the language-based procedural memory system.

In order to highlight the developmental aspects, we adopted the terminology “procedural learning system” and proposed that dyslexia could be assigned to the cerebellar branch of the language-based procedural learning system.

This eliminated (or at least finessed) the issue of which aspects of the network were actually the “culprit” and which the “bystander.”

There is strong, recent evidence for the framework, which is consistent with both automatisation deficit and cerebellar deficit (and provides a natural account of the phonological deficits).

For more specific evidence for the network analysis, serial reaction time studies (procedural learning) show a consistent deficit for dyslexia, coupled with consistent problems in procedural learning (Lum et al., 2013). Deficits in consolidation of procedural skill learning in dyslexic students have also been found (Nicolson et al., 2010).

Interestingly, there is also a greater impact on the procedural learning of letters than motor sequences (Gabay et al., 2012). Most intriguingly, a study has demonstrated better performance for dyslexic children than age-matched controls for learning and retention of declarative memory (Hedenius et al., 2013).

It is important, however, to acknowledge that more recent research has revealed the existence of many more neural networks than originally identified, as we discuss in the section “The Cognitive Neuroscience of Neural Network Development.”

Genetic Level

There is clear evidence of genetic transmission of dyslexia—a male child with dyslexic parent or sibling has a 50% chance of being dyslexic (Pennington et al., 1991).

There has been very extensive genetic research over the past 15 years, and genetic theories have identified a range of genes, many of which are involved in neuronal migration. Unfortunately, there has been a disappointing lack of progress, which contrasts markedly with the transformation in genetics techniques over that period, and the extensive research that has taken place (Carrion-Castillo et al., 2013; Becker et al., 2014).

A key difficulty is that genetic analyses cannot perform any better than the phenotypes (behavioral manifestations) collected, and given that reading difficulty is too diffuse a symptom, an appropriate phenotype or endophenotype is dependent on the quality of the theoretical framework investigated.

In short, genetic analyses are best suited to providing converging evidence relating to current theories, rather than directing the development of new theories.

The Cognitive Neuroscience of Neural Network Development

A major development in brain imaging research in the past decade has been the development of the tools to investigate structure and function at the network level. In particular, Diffusion Tensor Imaging (DTI) allowed the identification of white matter tracts and, in parallel, analysis of functional synchrony over time facilitated the identification of intrinsic connectivity networks.

Initial research led to the identification of the “Default Network (or Default Mode Network; Buckner et al., 2008), which is engaged when a person is not actively doing anything, and can be involved in thinking about others, thinking about themselves, remembering the past, and planning for the future.

Subsequent research (Yeo et al., 2011) highlighted a further six networks: the somatomotor network relates to the body and to motor coordination.

The dorsal attentional network is thought to mediate the top-down guided voluntary allocation of attention to locations or features (Vossel et al., 2014). The ventral attentional network is alternatively termed the Cingulo-Opercular network, and is often labeled the salience network.

The fronto-parietal network seems to initiate and adjust control; the cingulo-opercular component provides stable ‘set-maintenance’ over entire task epochs” (Dosenbach et al., 2008).

Early research limited network analysis to the cerebral cortex, but subsequent research established that the cerebellum was involved in all seven networks (Buckner et al., 2011) “Quantitative analysis of 17 distinct cerebral networks revealed that the extent of the cerebellum dedicated to each network is proportional to the network’s extent in the cerebrum with a few exceptions, including primary visual cortex, which is not represented in the cerebellum.” A valuable overview of the power of functional connectivity analyses in the case of autism spectrum disorder is provided by D’Mello and Stoodley (2015).

Converging evidence regarding the structural linkage at the systems level between key brain regions, namely the frontal cortex, the basal ganglia and the cerebellum has also recently emerged (Caligiore et al., 2017).

More recent analyses have investigated how these functional networks develop with maturation. A clear review of developments in early childhood is provided in Gilmore et al. (2018), who conclude (p. 134) that “Studies to date have found that, by birth, major white-matter tracts are in place and white-matter structural networks and sensorimotor resting-state functional networks are well developed. The first year of life is a period of robust gray-matter growth, rapid myelination and maturation of the microstructure of existing white-matter tracts and development of higher-order resting-state functional networks. By age 2 years, the fundamental structural and functional architecture of the brain seem to be in place, and the brain maturation that occurs in later childhood is much slower.” A recent study of development of attentional networks is provided by Rohr et al. (2018). In the case of reading, it is well known that the neural circuits involved show major structural changes with expertise, leading to the integration of the “Visual Word Form Area” (VWFA) into the initial circuitry (Schlaggar and McCandliss, 2007; Ben-Shachar et al., 2011). Naturally, given the lack of reading fluency for dyslexic children, there are clear differences in VWFA connectivity and function (van der Mark et al., 20092011; Koyama et al., 2013; Finn et al., 2014; Schurz et al., 2015). Unfortunately, it is not clear why these differences arise and how best to facilitate development of efficient connectivity in these cases.

In summary, the major recent development in cognitive neuroscience has been the identification of a range of neural networks that develop in early childhood. Unfortunately, almost all explanatory theories for dyslexia predate these insights, and so we now have the opportunity to revisit these theories in the light of these recent developments. We propose that the framework of learning and network development provides unique insights into the development of dyslexia, and indeed opportunities for dyslexia support, as we discuss below.

Georgetown University Medical Center
Media Contacts:
Karen Teber – Georgetown University Medical Center
Image Source:
The image is adapted from the Georgetown University news release.

Original Research: Open access
“Cerebellar function in children with and without dyslexia during single word processing”. Sikoya M. Ashburn, D. Lynn Flowers, Eileen M. Napoliello, Guinevere F. Eden.
Human Brain Mapping doi:10.1002/hbm.24792.


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.