Understanding the Interplay Between Executive Functions and Reading Development: A Challenge for Researchers and Practitioners Alike
In June of 2022, The Dyslexia Foundation (TDF) organized a convening of dyslexia researchers and practitioners around the topic of executive functions. There was consensus on the importance of executive functions for reading development. However, the difficulty of defining, measuring, and training executive functions emerged as a challenge for researchers and practitioners alike. The upcoming special issue of Mind Brain and Education presents a collection of articles that survey different perspectives, define the current knowledge base, highlight challenges and inconsistencies in research, and chart a path towards a more nuanced understanding of the role of executive functions in reading and dyslexia.
In June of 2022, The Dyslexia Foundation (TDF) organized a convening of dyslexia researchers and practitioners around the topic of executive functions (EF). Researchers and practitioners alike reached consensus that EF are important for reading and play a role in dyslexia. While this conclusion might have been novel and even controversial a decade ago, the idea that EF plays a role in dyslexia is now commonplace in the literature and most dyslexia researchers are likely to endorse a link between EF and reading skills. For example, there are at least three recent meta-analyses that conclude a moderate but reliable relationship between EF and reading development (Lonergan et al., 2019; Peng et al., 2022; Spiegel, Goodrich, Morris, Osborne, & Lonigan, 2021). However, even though we might surmise that the existence of a link between EF and reading ability is clear, it is not clear what exactly is meant by the term EF. Researchers and practitioners both use the term EF in reference to a heterogeneous mix of scientific measures and real-world observations. This leaves us with a conundrum: we know that EF is important but we cannot precisely define it or measure it.
The EF conundrum can be appreciated by the simple observation that some articles refer to “executive function” in the singular, suggesting a unitary construct, whereas others refer to “executive functions” in the plural, suggesting that the construct consists of a plurality of distinct yet related skills (here we use the same abbreviation EF for singular and plural). Needless to say, a construct that is hard to define is also hard to measure, a challenge for researchers and practitioners alike. This conundrum emerged early in the TDF convening and can be appreciated in the collection of articles in this special issue. Dyslexia researchers are reaching consensus that EF is important, yet, defining what it is, how it should be measured, and the mechanisms by which it influences (and/or is influenced by) reading skills are far from clear. Similarly, dyslexia practitioners are well aware that individual differences in EF contribute to learning challenges but practitioners struggle to define what exactly is encompassed by EF and how to best support EF in children with dyslexia. Thus, understanding the interplay between EF and reading development is an important scientific challenge with far-reaching implications for general education, special education, and private schools serving children with learning disabilities.
EXECUTIVE FUNCTIONS FROM THE PERSPECTIVE OF BEHAVIORAL MEASURES
Cirino (this issue) spotlights the EF conundrum by both providing a working definition of EF and also highlighting the inconsistencies between theory and data. For example, if we take a theoretical approach to defining EF as “a complex higher-order construct supporting goal-directed behavior,” then we can delineate various high-order constructs that are theoretically important for reading: deriving meaning from a sequence of words requires a controlled process to hold different ideas in working memory and draw links between them; recognizing a word with an irregular spelling pattern requires inhibiting the phonemes that are typically associated with a letter or resolving conflict between different potential pronunciations; learning to read in an opaque orthography requires flexibly switching between different rules governing the probabilistic relationship between orthography and phonology. Although one can take a theoretical approach to outline different high-order cognitive processes that are required for fluent reading, a scientific theory must make falsifiable predictions (Popper, 1959). Even if we start from theoretical principles, we must immediately confront measurement challenges as we try to operationalize EF into surveys and tasks. Cirino then raises one of the fundamental measurement challenges for EF: measurement impurity. Executive functions are unobservable latent constructs that do not have a direct measure. While this is arguably true about all cognitive constructs, the measurement impurity problem is much more dramatic for EF than other domains. For example, the myriad of measures that tap into decoding ability are all highly correlated. Timed and untimed measures of real word and pseudoword decoding are generally correlated between r = 0.8 and r = 0.95, making it straightforward to design and validate a new decoding measure (Yeatman et al., 2021). The myriad of measures that are purported to tap into EF, on the other hand, are only moderately correlated at best. Even if we are careful to define separate components of EF such as inhibitory control, working memory, and cognitive flexibility (or shifting), correlations as low as r = 0.3 are not uncommon for different measures of the same component process (Cirino et al., 2018; Eisenberg et al., 2019; Friedman & Miyake, 2017; Karr et al., 2018; Miyake & Friedman, 2012). Thus, testing the hypothesis that, for example, inhibitory control is related to reading skills is difficult because there are dozens of proposed measures of inhibitory control that are only loosely correlated with each other. A correlation between any particular measure and reading ability cannot, therefore, be attributed to the construct of inhibitory control because the relationship might reflect other performance limiting factors. Factor analysis and structural equation modeling can be used to mitigate this problem but, in practice, it is not feasible to collect a sufficient number of different measures to solve the task impurity problem for any component of EF.
This illustrates that, while task impurity is a challenge in all areas of psychological science, it is particularly pronounced in the study of EF where the average correlation between tasks tapping into the same construct is particularly low. For many widely used EF tasks (e.g., Flanker) only 10%–20% of the variance in task performance reflects the particular EF construct. Thus, developing a theoretical framework linking EF with reading development might be straightforward, but an empirical test is much more difficult.
EXECUTIVE FUNCTIONS FROM THE PERSPECTIVE OF BRAIN IMAGING MEASURES
Church-Lang (this issue) provides an empirical example of the challenges raised by Cirino (this issue). A series of fMRI studies relating the neural underpinnings of EF to reading ability are introduced. For example, Roe et al. (2018) defined regions of the “frontoparietal control network” which is often described as a domain-general network involved in cognitive control. They found that individual differences in the fMRI response amplitude within this network varied during a reading comprehension task and were correlated with measures of reading ability. However, fMRI responses in the same regions during the stop signal task—a classic response inhibition task—did not correlate with reading ability. This information prompts the question, “how domain-general is this network?” If the functions of this brain network truly cut across tasks, then one would expect to see correlations with reading ability across a myriad of different tasks.
Banich and colleagues (this issue) define an approach to identify brain regions that should be considered “domain-general” and then use this approach to investigate the link between domain-general neural processes and reading comprehension. They reason that a domain-general brain region involved in reading should show activation during reading as well as other academic functions such as math and domain-general cognitive processes such as working memory. They identify a sequence of domain-general regions in the inferior frontal lobe that meet this criteria—overlapping activation across tasks—and then investigate the relationship between individual differences in activation and reading achievement. In terms of activation levels and functional connectivity, Banich and colleagues show robust relationships between the function of domain-general brain regions and reading development. Thus, even though more work needs to be done to understand the precise role that these domain-general neural processes play in reading development, work by Banich and others has demonstrated the importance of understanding these brain networks for a more complete model of reading development.
Burgess and Cutting (this issue) provide another example of how a deeper understanding of EF is critical to more nuanced models of reading development. Despite the challenges of reaching a consensus definition of EF, they posit that “there is little question to date as to whether EF is linked to reading; instead, the question is how is EF linked.” From a theoretical perspective, EF has relevance for all the most widely cited theoretical models of reading and an average effect size of 0.3 based on recent meta-analyses (Follmer, 2018; Spiegel et al., 2021). Burgess and Cutting leverage longitudinal data and structural equation modeling to examine the dynamics between EF, language, decoding, and reading comprehension skills. This sequence of studies provides support for the “behind the scenes” role that different EF components play in supporting various aspects of reading development (Aboud, Barquero, & Cutting, 2018; Spencer, Richmond, & Cutting, 2020). More specifically, they assert an indirect relationship whereby executive functions influence decoding skills which, in turn, influence reading comprehension. To confront the measurement challenges raised by Cirino (this issue), Burgess and Cutting define an elegant methodology to combine behavioral and neuroimaging methodologies to unpack the way that different brain networks are involved in coordinating complex processes like reading.
Yamasaki and Prat (this issue) provide complementary perspectives on how measuring neural processes such as efficiency, adaptability, and synchronization can further shed light on the neural underpinnings of EF. Unpacking the causal relationships between specific components of EF and how they influence the myriad of factors associated with skilled reading will remain an important challenge for future research.
The collection of articles in this special issue provide a framework for the field to continue making progress clarifying the construct of EF through the combination of complementary measurement modalities spanning behavioral tasks and observations, neuroimaging, and environmental factors.
EXECUTIVE FUNCTIONS AND SOCIOECONOMIC STATUS: THE INTERPLAY BETWEEN ENVIRONMENT, BRAIN DEVELOPMENT AND LEARNING
Conventionally, many researchers ask the question of how EF “deficits” relate to academic outcomes like reading. Taylor, Abdurokhmonova and Romeo (this issue) urge us to consider EF in terms of adaptations to the environment as opposed to cognitive deficits. Children raised in poverty face a myriad of challenges and overcome these challenges through adaptations that are often overlooked by researchers. These “adaptations to adversity” (Taylor et al., this issue) lead to different strategies for confronting various tasks designed by researchers (Miller-Cotto, Smith, Wang, & Ribner, 2022). This fact further complicates the measurement challenges noted in the previous section and highlights the importance of understanding the various factors that contribute to differences in task performance other than “deficits in EF.” For example Taylor et al. lay out a body of evidence showing that the neurobiological foundations of individual differences in EF are qualitatively different for children raised in high-SES versus low-SES environments (Romeo et al., 2017, 2022; Ursache et al., 2016; Ursache & Noble, 2016). Thus, the same EF task is likely tapping into different neural systems for different children, and these differences can only be understood through careful consideration of the interplay between brain development and environmental factors. One of the many important environmental considerations is the impact of stress hormones on brain development. This is a point that is investigated through a randomized controlled trial by Buchweitz et al. (this issue) to understand how reading and anxiety are differentially related to the release of stress hormones in children of different reading levels. Another important environmental factor that is investigated in this issue is the effect of exposure to pollution on brain development, EF, and reading development (Margolis and Greenwood, this issue). Similar to the perspective of Taylor et al., Margolis and Greenwood propose EF as a mediator in the relationship between environmental factors associated with poverty (e.g., higher exposure to pollution) and academic outcomes such as reading. Together, these papers prompt us to reconsider our notions of EF within the context of understanding children’s adaptations to environmental differences.
CONCLUSIONS
The collection of papers in this special issue all demonstrate different ways that EF are related to reading development. They exemplify different approaches and challenges to measurement—spanning behavioral, environmental, and neural measures—and represent different theoretical approaches to operationalizing EF and the link with reading. Taken together, this collection of papers can be seen as an introduction to both the promise and challenge of understanding the link between EF and reading development. No paper disputes the fundamental importance of EF, yet each paper provides a different perspective on how dyslexia researchers and practitioners should be integrating EF into their working model of reading development.
Acknowledgments
I would like to thank The Dyslexia Foundation for a convening that bridged dyslexia research and practice along with the exceptional presenters that contributed to this special issue. I would also like to thank all the schools that attended The Dyslexia Foundation meeting and helped keep research grounded in issues of practical importance. This work was funded by NICHD R01HD095861.