This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Main Theoretical Question Eye movement patterns during reading in normal skilled readers are highly complex and systematic. Our main theoretical question is how do these patterns arise? And in particular: How do the patterns emerge from an algorithm which is believed to be implemented in the brain? Our hypothesis is that they arise through a combination of task demands and perceptual, cognitive and motor constraints to which each reader is subject. To investigate this hypothesis, we are studying the development of reading behavior in a computational learning system based on Reinforcement Learning Background This research project uses a computational model to investigate the ways in which various cognitive (i.e. high-level mental) processes interact with low-level nervous system processes in order to exhibit the eye-movements seen in reading and visual search behavior. It is at the intersection of Computer Science, Neuroscience and Psychology since it investigates how a biologically-inspired but resource-intensive algorithm called Reinforcement Learning gives rise to intelligent human-like reading behavior. Eye movements are controlled by a variety of brain regions and because of this dependence, it is not surprising that eye movements are used to diagnose a wide range of disorders, since their disturbance can pinpoint specific locations where brain trauma is likely to have occurred. One of the regions which appears to have a part in at least learning (if not also executing) eye movement strategies is the basal ganglia. An important part of the basal ganglia the substantia nigra pars compacta contains neurons that secrete a neurotransmitter called Dopamine, and is subject to neurodegeneration in Parkinson s disease. Current thinking in computational neuroscience is that the basal ganglia, and specifically the Dopamine neurons of the substantia nigra, are a part of the brain that is critical for Reinforcement Learning. Because of their resource-intensive nature, specific subgoals of our research project that will benefit from Pittsburgh Supercomputer Center resources are 1) to increase the realism of the simulations that we have made to date for quantitative testing of the model using time steps of finer granularity; 2) to investigate the effects of language processing systems on eye movement patterns, and, 3) to investigate how alternative attention resource allocation schemes influence eye movements, since it is known that eye movements and visual attention are independent. All three of these objectives increase the RAM requirements in our simulations because they a) increase the complexity of each state object, thereby increasing its size, and/or b) increase the number of state objects that must be tracked. We are specifically interested in pursuing item (3) above in this initial grant, and items (1) and (2) as there are remaining service units available. Eye Movements During reading, it is commonly believed that one s eyes move smoothly across the words in the text, occasionally pausing and/or returning to difficult or confusing words. However, it is now well known that eye movements during reading are made up of a large number of rapid, discrete movements call
Showing the most recent 10 out of 292 publications
