Modeling Human Cognitive Processes for Reasoning in LLMs
Event details
| Date | 26.08.2025 |
| Hour | 17:00 › 19:00 |
| Speaker | Marta Knezevic |
| Location | |
| Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Antoine Bosselut
Thesis advisor: Prof. Tanja Käser
Co-examiner: Prof. Martin Schrimpf
Abstract
This thesis explores learning by interaction and in-context learning in humans and LLMs. We investigate how large LLMs can be guided toward more human-like reasoning by aligning their learning and inference processes with cognitive science and educational theory. Drawing from principles such as metacognition, analogical reasoning, and strategic behavior, the research investigates how human-inspired concepts can improve LLMs' ability to do inductive reasoning, hypothesize, learn from mistakes and reflect on their own outputs. This work aims to bridge the gap between human and machine learning and uncover shared principles of representation, abstraction, and reasoning that drive intelligent behavior in both
Selected papers
coming soon
Exam president: Prof. Antoine Bosselut
Thesis advisor: Prof. Tanja Käser
Co-examiner: Prof. Martin Schrimpf
Abstract
This thesis explores learning by interaction and in-context learning in humans and LLMs. We investigate how large LLMs can be guided toward more human-like reasoning by aligning their learning and inference processes with cognitive science and educational theory. Drawing from principles such as metacognition, analogical reasoning, and strategic behavior, the research investigates how human-inspired concepts can improve LLMs' ability to do inductive reasoning, hypothesize, learn from mistakes and reflect on their own outputs. This work aims to bridge the gap between human and machine learning and uncover shared principles of representation, abstraction, and reasoning that drive intelligent behavior in both
Selected papers
coming soon
Practical information
- General public
- Free
Contact
- edic@epfl.ch