Open Postdoctoral position, faculty mentor Jason Yeatman

Deep Phenotyping of Learning Differences

The high-level goal of this project is to understand the mechanisms underlying learning differences such as dyslexia and to do so in a large, representative sample through the ROAR platform. Excerpt from Specific Aims of the R01 funding this position are below.

Deep Phenotyping of Dyslexia Subtypes

Learning to read depends on a confluence of foundational skills. Some skills, such as phonological awareness, are known to be critical. However, the role of other factors – such as differences in visual processing and executive functions (EFs) – are still debated. The overarching goal of this project is to characterize consistency and heterogeneity in mechanisms associated with word reading difficulties (i.e., developmental dyslexia) using a deep phenotyping approach. For example, despite 40 years of controversy surrounding the relationship between visual deficits and developmental dyslexia, the field is no closer to resolving the debate over the explanatory power of rapid visual processing, motion perception or crowding in predicting reading skills. In regards to EF, the field is progressing towards a consensus that EFs are important for reading and that EF deficits are common in dyslexia. But due to the limited number of measures used in any one study, many questions about the relationship between EFs and dyslexia remain unresolved. Here we take an innovative approach involving a) development of new, open-source technology and b) large-scale data collection in a diverse sample spanning over 250 schools across 30 states to answer three significant questions regarding the mechanisms of word reading difficulties such as dyslexia:

• Do individual differences in visual processing and executive functions (EFs) explain additional variance in reading abilities above and beyond phonological awareness?
• Within the domains of visual processing and EF, does one factor or multiple independent factors contribute to differences in reading ability?
• Are measures of visual processing and EF useful additions to conventional dyslexia screening in kindergarten and first grade?

These questions represent long-standing scientific challenges with significant implications for dyslexia screening and intervention. We address these questions through an innovative methodology including:

• Open-source assessment platform: We have developed an online platform for reading assessment - the Rapid Online Assessment of Reading (ROAR) - that allows us to collect reliable and valid measures of reading ability, phonological processing, rapid automatized naming, visual processing, and executive functions at scale in diverse and representative samples. Further development of this open-source technology opens the door to future research directions and integration into educational practice.
• Research Practice Partnerships: Through partnerships with a) schools serving children with learning disabilities, b) charter schools, and c) large public schools, we will collect the first large-scale, longitudinal dataset with detailed measures of visual processing and EFs alongside measures of reading ability in a diverse sample of children with dyslexia and typical readers.
• Scientific reproducibility and transparency: Our team has a long history of developing open-source software to support rigorous, robust, and reproducible data analysis. Conventional statistical approaches will be combined with innovations in interpretable machine learning to address each aim from multiple angles. Analysis code will be distributed as open-source software to ensure reproducibility and transparency as well as supporting the extension of our approach to new domains.