Improving reading skills based on a better understanding of the neuronal underpinnings of visual processing in reading.

• Applicant: Professor Dr. Benjamin Gagl
• Subject Area: General, Cognitive and Mathematical Psychology ; Human Cognitive and Systems Neuroscience
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 523332674

The ability to read is a precondition for academic achievement and societal participation. Cognitive psychology provides several computationally explicit reading models that describe step-by-step and human-understandable core algorithms and mental representations involved in visual word recognition. Such models lead to empirically grounded applications, e.g., in the remediation of children's reading problems. However, current cognitive models implement underspecified visual processing, as the process is regarded as non-problematic and easy to implement. In data science, computer vision models implement sophisticated visual processing to identify letters and words from images. However, these models cannot be used to motivate treatment programs as model complexity is high. Thus, our understanding of the exact inner workings of these models is restricted. In this project, we plan to implement human-understandable visual processing from image-to-letter identification in the speechless reader model so that identification is accurate and font invariant. Conceptually, the model is a neuro-cognitive visual word recognition model. The speechless reader transparently, i.e., step-by-step and human-understandable, simulates visual word recognition behavior based on mechanistic neuronal and cognitive hypotheses. After implementing visual processing in the model, we plan to evaluate the new model assumptions based on electrophysiological brain measurements in response to words with different fonts. Finally, we plan a training study that will be motivated by the core assumptions of the novel model parts. This study will investigate if one can train visual processes specific to reading to increase reading skills in German language learners. Ultimately, we plan to have an evaluated visual word recognition model capable of simulating word recognition behavior accurately and transparently based on a word image input and a training procedure that helps readers with low skills.

• DFG Programme: Research Grants