Nora's PhD work focused on how the brain dynamically constructs sentence-level meanings from word-level features. Her research uses neural networks for Prediction and Interpretation of fMRI images of sentences. The neural network is trained to map word features (using brain-based attribute representations, Binder et al., 2009) to word fMRIs. Then, the FGREP neural network (Miikkulainen, 1991) is used to determine how the representations would have to change to predict the fMRI patterns more accurately. These changes represent the effect of context. The new word representations are multimodal; grounded in different brain systems. The context-based representations could be used for AI applications such as service robots. Her PhD research was a joint collaboration between UT Austin and ITESM, Mexico, with Risto Miikkulainen as her main advisor.