Researchers have developed a novel AI approach to understand decision-making processes in humans and animals. Traditional models often assume optimal decision-making based on past experiences, potentially missing real-world behaviors. A study led by scientists from New York University and the University of California, San Diego, employs tiny artificial neural networks to reveal how decisions are actually made.
Marcelo Mattar, an assistant professor at NYU’s Department of Psychology and one of the study’s authors, explained: “Instead of assuming how brains should learn in optimizing our decisions, we developed an alternative approach to discover how individual brains actually learn to make decisions.” This method uses small neural networks that are powerful enough to capture complex behavior yet simple enough for analysis.
The research suggests these smaller networks can predict choices more accurately than classical cognitive models by highlighting suboptimal behavioral patterns. Ji-An Li, a doctoral student at UC San Diego, emphasized the advantage of using small networks: “They enable us to deploy mathematical tools to easily interpret the reasons behind an individual’s choices.”
Marcus Benna from UC San Diego noted that while large AI models excel at prediction tasks like movie recommendations, understanding their strategies is challenging. The simpler AI models used in this study allow researchers to analyze decision-making dynamics more clearly.
The model successfully matched decision-making processes across species and predicted suboptimal decisions reflective of real-world scenarios. Mattar highlighted the broader implications: “Just as studying individual differences in physical characteristics has revolutionized medicine, understanding individual differences in decision-making strategies could transform our approach to mental health and cognitive function.”
This research was supported by grants from various organizations including the National Science Foundation and UC San Diego’s California Institute for Telecommunications and Information Technology/Qualcomm Institute.
