Adam Morgan

@adumbmoron.bsky.social

865 Followers 439 Following

Postdoc at NYU using ECoG to study how the brain translates from thought to language. On the job market! 🏳️‍🌈🏳️‍⚧️🗳️ he/him https://adam-milton-morgan.github.io/

Evaluating the LLM's, benepar's, and the Stanford Parser's VFF estimates by comparison to Gahl et al.'s (2004) database. The LLM produced the best fit, across 7 different verb frames.
Accuracy for the GPT-4o (LLM) parser, Berkeley Neural Parser (benepar), and Stanford Parser on three manually-parsed verbs. The LLM consistently showed higher agreement with manual parses.
Density plots for the number of detections of subjects (left) and objects (right) during the production of subjects and objects in passive sentences, split by two prefrontal regions: IFG (top) and MFG (bottom). IFG sustained representations of subjects throughout both words while MFG sustained representations of objects.
Decoding results from middle frontal gyrus during passive sentences showed sustained encoding of the object noun.
Decoding results from sensorimotor cortex for active sentences: the subject noun is predicted above chance while it is being said, and the object noun while it is being said.
Word-specific patterns of neural activity: electrodes that selectively responded to each of the six words.
Task screenshots (picture naming: a cartoon picture of Frankenstein; scene description: cartoon image of Dracula hitting Frankenstein) and mean neural activity per word for one electrode in middle temporal gyrus.
Towet Village, Papua New Guinea
Results of decoding words during the production of active and passive sentences. In actives, nouns were decoded in the order they were said, whereas in passives, prefrontal cortex sustained representations of both the subject and the object throughout the duration of the sentence while sensorimotor areas patterned with actives (showing “congruent” temporal representations).
Barplot showing that nearly all incongruent detections (i.e., prediction of the subject while patient said the object or vice versa) came from passive sentences. Pie charts show where these detections were made. The majority came from prefrontal cortex, with inferior frontal gyrus (IFG) representing the subject and middle frontal gyrus (MFG) representing the object.
Results from using the same classifier in the previous image (trained on picture naming data from sensorimotor cortex) to decode word identity throughout the production of passive sentences like "Frankenstein is being hit by Dracula." As before, the subject was detected when the patient said the subject, and the object when the patient said the object.
Prediction accuracy over time for the subject and object of a sentence. Classifier was trained on picture naming data from 50-100ms after speech onset and tested at each time point throughout sentence production. It significantly predicted the subject (Dracula, in the example sentence) when the patient said the subject and the object (Frankenstein) when the patient said the object.
Unique cortical activity patterns for each of the 6 words (chicken, dog, Frankenstein, Dracula, ninja, nurse).
Depiction of a picture naming trial (patient sees Frankenstein and says "Frankenstein"), a sentence production trial (patient sees Dracula hitting Frankenstein and says "Dracula hit Frankenstein"), and mean neural activity for each of the six words in one electrode during picture naming.