Michael Hahn

Michael Hahn is a Tenure-Track Professor (W2) at Saarland Informatics Campus, Saarland University, where he directs the Language, Computation, and Cognition Lab (LaCoCo). He is affiliated with the Departments of Language Science and Technology and Computer Science, and holds additional distinctions as an ELLIS Member and ELIZA Fellow.

Dr. Hahn received his PhD from Stanford University in 2022, advised by Judith Degen and Dan Jurafsky. His academic journey has positioned him at the forefront of research bridging artificial intelligence, cognitive science, and linguistics, with a focus on understanding language models and human cognition through computational approaches.

His research interests include:

• Information Processing in LLMs and the Brain
• Abilities of Language Models and how they are learned through theoretical and empirical methods
• Understanding Inner Workings of Language Models via mechanistic reverse-engineering
• Computational Cognition and Neuroscience, modeling human language processing and perception

Dr. Hahn's publications reveal a strong emphasis on theoretical foundations of language models, with significant contributions to understanding transformer architectures, language model capabilities, and their relationship to human cognition. His work spans from establishing formal frameworks for length generalization to developing methods for neural activation analysis, consistently appearing in top-tier venues including ICML, ICLR, ACL, and NeurIPS.

His scientific recognition includes:

• Best Paper Award at ACL 2024 for "Why are Sensitive Functions Hard for Transformers?"
• Second Prize at ICML 2024 Mechanistic Interpretability Workshop for "InversionView for reading out information from neural activations"
• Recognition as Outstanding Area Chair at EMNLP 2024

Dr. Hahn actively mentors students and researchers, supervising PhD students, postdoctoral researchers, and research assistants. His lab has demonstrated remarkable productivity, with multiple papers accepted at premier conferences including three at NeurIPS 2024 and works at ICLR 2025 and ICML 2025. He is also involved in the academic community through co-organizing workshops and delivering invited talks.

The Language, Computation, and Cognition Lab (LaCoCo) serves as a hub for cutting-edge interdisciplinary research, investigating the fundamental connections between machine learning, language, and the human mind, with a particular emphasis on understanding the relationship between artificial and human intelligence.