In Fall 2020, students were forced to take a remote version of the course due to the Covid19 pandemic. Unfortunately, this meant that students were unable to design and run their own original study. Instead, students developed original research questions and conducted secondary data analysis projects using data that was publicly available on OpenNeuro. One student worked with the naturalistic Paranoia dataset Finn et al., 2018. The other group worked with a dataset in which participants underwent three separate scanning sessions to evaluate brain responses to pictures of social interactions, food, or flower following a 10-hour food deprivation, social-isolation, or baseline manipulation Tomova et al., 2020.
Nouns and verbs, adjectives and adverbs: An investigation into syntactical localization using fMRI#
Part of speech analysis forms an important part of linguistic, psychological, and neuroscientific studies of language. A wide variety of studies have shown various active regions when discriminating between nouns and verbs, but few have looked at naturalistic paradigms. Using the dataset provided in Finn et al. (2018), I examined the neural correlates of nouns, verbs, adjectives, and adverbs. Through univariate contrast analysis, I found preferential activation for nouns bilaterally in the posterior portion of the inferior temporal gyrus, as well as the left superior temporal gyrus and angular gyrus. Verbs activated the bilateral temporal poles, as well as the cerebellum. Adjectives showed increased activation in the right hemisphere, while adverbs showed few preferred regions. Nouns and verbs together showed strong activation in the left superior temporal gyrus as compared to adjectives and adverbs. Using Multivariate Pattern Analysis (MVPA), my Support Vector Machine (SVM) was able to predict part of speech with 80-84% accuracy. These findings provide promising evidence that part of speech can be localized using naturalistic language paradigms.