Periodic Reporting for period 1 - ROAR (Investigating the Role of Attention in Reading)
Okres sprawozdawczy: 2019-10-01 do 2021-09-30
In total, the project comprised five objectives:
- To reveal word-to-word influences during reading, such as influences of the syntactic categories of words on the ease of recognizing adjacent words.
- To determine the distribution of attention during reading (our findings suggest that multiple words are processed in parallel, at relatively 'deep' levels (i.e. not just letter information, but also information about grammar).
- To determine how the brain computes word order (e.g. in the phrase "Do love you me?" many readers do not notice the incorrect word order. Our research suggests that this is due to grammatical constraints that are in play while readers process multiple words simultaneously).
- To find out whether dyslexia may be caused by attentional problems.
- To summarize current knowledge about the reading process in a computational model - that is, a computer implementation of our current theoretical framework, called OB1-reader.
In short, we established that during reading, your attention extends beyond the word that you're looking at. This means that while you're looking at (and trying to recognize) a word, you're (either consciously or sub-consciously) also processing information from surrounding words. These influence the recognition process. We found that this widespread attention is rather innate: it is depicted even by children who have just learned how to read.
In other studies, we used the notion of widespread attention to develop an interface intended to improving reading in patients with macular degeneration. This interface appears to work really well in participants for whom we simulated a foveal scotoma (blind spot) using eye-tracking technology. That is, we have a successful proof of concept in visually healthy participants for whom we simulated the disease. The efficacy of our interface is still to be tested with actual patients.
Various experiments have guided the ongoing development of our model of reading, OB1-reader. This computational model (i.e. it simulates reading behavior (eye movements, word recognition speed) for a given text input) is to help us understand how reading works in the brain. The more the simulations mimic what we observe in actual humans, the more evidence we have for the model's underlying assumptions.