IC Colloquium : The complex brain representations involved in natural reading
Event details
| Date | 02.02.2015 |
| Hour | 10:15 › 11:30 |
| Location | |
| Category | Conferences - Seminars |
By : Leila Wehbe - Carnegie Mellon University
IC Faculty candidate
Abstract :
How is information organized in the brain during natural reading? Where and when do the required processes occur, such as perceiving the individual words and combining them with the previous words? How are different types of information represented, such as semantics, syntax or higher-level narrative structure? Due to the complexity of language, most brain imaging studies have concentrated on one aspect of language processing, and usually use highly controlled stimuli. Such artificial stimuli might lead to conclusions that do not generalize beyond the experimental setting. My research revolves around studying the complex parallel process involved when participants read a natural text in a close to normal setting.
My approach uses natural language processing to model the content of the stimulus text, and machine learning tools to relate it to the activity of different brain regions and make inferences on their roles. I will describe results from an fMRI reading experiment that allowed us to construct a map detailing how different regions of the brain participate in different reading sub-processes. I will also describe a Magnetoencephalography study that suggests a time-line for how meaning is built and updated while reading. This approach is broadly applicable to studying higher-level functions, and could help us understand individual differences in behavior by relating them to differences in the organization of information in the brain. It might also enable us to build better statistical language models that incorporate brain imaging data in their learning phase.
Bio :
Leila Wehbe is a PhD student in Machine Learning at Carnegie Mellon University. She is advised by Tom Mitchell. Her research interests include the neurobiology of language, naturalistic brain imaging, combining natural language processing with brain imaging of language and nonparametric hypothesis testing methods. She received her BE in Electrical and Computer Engineering from the American University of Beirut.
More information
IC Faculty candidate
Abstract :
How is information organized in the brain during natural reading? Where and when do the required processes occur, such as perceiving the individual words and combining them with the previous words? How are different types of information represented, such as semantics, syntax or higher-level narrative structure? Due to the complexity of language, most brain imaging studies have concentrated on one aspect of language processing, and usually use highly controlled stimuli. Such artificial stimuli might lead to conclusions that do not generalize beyond the experimental setting. My research revolves around studying the complex parallel process involved when participants read a natural text in a close to normal setting.
My approach uses natural language processing to model the content of the stimulus text, and machine learning tools to relate it to the activity of different brain regions and make inferences on their roles. I will describe results from an fMRI reading experiment that allowed us to construct a map detailing how different regions of the brain participate in different reading sub-processes. I will also describe a Magnetoencephalography study that suggests a time-line for how meaning is built and updated while reading. This approach is broadly applicable to studying higher-level functions, and could help us understand individual differences in behavior by relating them to differences in the organization of information in the brain. It might also enable us to build better statistical language models that incorporate brain imaging data in their learning phase.
Bio :
Leila Wehbe is a PhD student in Machine Learning at Carnegie Mellon University. She is advised by Tom Mitchell. Her research interests include the neurobiology of language, naturalistic brain imaging, combining natural language processing with brain imaging of language and nonparametric hypothesis testing methods. She received her BE in Electrical and Computer Engineering from the American University of Beirut.
More information
Practical information
- General public
- Free
- This event is internal
Contact
- Host : Jim Larus