Amirhossein Mohammadi

I am a Ph.D. candidate at Utrecht University, supervised by Robert Bagheri, Albert Gatt, and Laurence Frank, working at the intersection of computational linguistics, cognitive science, and artificial intelligence. My research investigates the capabilities and limitations of large language models in computing and acquiring natural and unnatural languages. I am particularly interested in understanding how LLMs process, recover, and interpret structurally degraded input, and what this reveals about the relationship between neural architectural biases and human cognitive constraints on language.

My broader research interests include mechanistic interpretability of language models, information locality in natural language, and the application of NLP methods to mental health.