Emotion and Cognitive Models
Emotions play a powerful, central role in everyday life. They shape how we perceive the world, bias our beliefs, influence our decisions and in large measure guide how we adapt our behavior to the physical and social environment. We have been developing a a range of computational models of emotion’s influence on cognition. The work began with early work on the CBI’s (Carmen’s Bright Ideas) seminal model of coping strategies. CBI began the effort to formalize how emotion, specially an apparaisal based model of emotion, influenced coping strategies which in turn influenced cognition (Marsella et al, 2000, 2003). This work integrating appraisal and coping continued in the work on EMA (Emotion and Adaptation) (Gratch and Marsella 2001; Marsella and Gratch 2003; Gratch and Marsella 2004), culiminating with work that formalized clinical reseacrh on coping in terms of changes to attention, beliefs, desires and intentions (Marsella and Gratch, 2009). This work was evaluated in a range of laboratory studies that assessed the ability to model human emotional behavior. These works were also applied to various training scenarios that incorporated the models into embodied conversational agents or virtual humans. More recent work has shifted to using models of emotion and coping in social simulations of real world scenarios that impact socio-political policies, such as hurricane evacuation (Yongsatianchot & Marsella, 2022). Also in keeping with the explosion of work using Large Language Models (LLMs), we have also explored the degree to which representation of emotion and coping phenomena in LLMs is aligned with psychological theories and empirical data on emotion related human behavior (Yongsatianchotet al. 2023,2024).
Expert Decision Making Under Risk and Uncertainty
Experts often make decisions in environments where information is incomplete, ambiguous, or rapidly changing. These high pressure scenarios, whether in healthcare, supply chain management, or other fields, demand not only deep knowledge but also the ability to effectively navigate uncertainty and stress. The ability to make timely, effective decisions is essential, but even the most skilled professionals face significant challenges when confronted with uncertainty, stress, and the need for rapid adaptation.
Unlike traditional models that treat decision making as a static process, we aim to capture the dynamic nature of expert decision making in real time. By creating attention driven computational agents, we simulate how experts process information, manage uncertainty, and adapt their strategies as conditions change. One key application of our work is in the pharmaceutical supply chain, where we are designing models based on the insights of real world experts to support decision makers by providing insights into how they can navigate uncertainty.
Our research is part of a broader effort to build computational models that reflect the real world challenges faced by experts. We focus on how experts cope with stress, uncertainty, and high pressure environments, aiming to enhance the quality and efficiency of their decisions. By advancing these models, we are developing decision support systems for the most complex and uncertain situations.
Modeling Cognitive Biases
Cognitive heuristics are mental shortcuts that help individuals make decisions efficiently, allowing them to navigate daily challenges while conserving cognitive energy. While these heuristics are often useful, efficient biases that address bounds on rationality, the systematic deviations from rational decision-making that arise from these shortcuts can lead to flawed judgments and costly errors.
In this research, we focus on computationally modeling cognitive biases that simulate how these biases influence human decision-making. These models can inform the design of collaborative and competitive AI agents, enabling them to better understand the extent to which a human collaborator or adversary may be vulnerable to specific biases. By incorporating this understanding, agents can adopt more strategic approaches—whether by helping to address biases in a collaborative context to enhance teamwork, or by leveraging them in competitive scenarios to gain an advantage.