On July 3, 2026, researchers from the Center Synergy of Systems at TUD Dresden University of Technology, along with the Max Planck Institute for Human Development and the University of Basel, introduced a novel approach to understanding human decision-making. Their findings, published in the Proceedings of the National Academy of Sciences, reveal how free-text answers combined with large language models (LLMs) can unveil the hidden reasons behind human choices.
Exploring Human Decision-Making
Understanding why individuals make specific choices has long been a challenge in psychology and behavioral science. According to lead author Dr. Kamil Fuławka, a researcher at SynoSys, asking participants to elaborate on their decision processes can enhance insights into human behavior. The research team combined behavioral experiments with participants' self-reported explanations to create a comprehensive methodology for analyzing decision-making.
In their experiment, participants engaged in gambling activities and were tasked with explaining their decisions in their own words. The researchers developed a framework that utilized existing theories of decision-making to categorize potential reasons for decisions, such as maximizing gains or minimizing losses.
Leveraging Large Language Models
The study highlighted the effectiveness of using large language models to analyze free-text data. By identifying common themes and decision reasons within the participants' explanations, the researchers were able to validate their findings through mathematical modeling of the choices made. This integration of qualitative and quantitative data presents a significant advancement in behavioral research.
