1. Simulated virtual worlds and hybrid societies
Emerging AI systems are reshaping collective behavior and our society. As humans and AI agents interact within shared networks, their influence on one another can be nonlinear and difficult to observe with traditional methods. We build virtual worlds and programmable social networks that let us control the rules of interaction and study large groups in a controlled setting. Our research, supported by two NSF awards, examines how collectives become more creative and intelligent—and how hybrid societies of humans and AI evolve.
Related publications
2. Internal representations
Human perception is rich, multi-dimensional and contextual. (Consider, for example, the ways in which emotion is conveyed by the voice: by pitch, volume, and many other parameters.) Yet behavioral methods, biased by their limitation to one-dimensional and simplified stimulus spaces, typically produce an impoverished understanding of human perception. Inspired by Monte Carlo Markov Chain techniques borrowed from machine learning and physics, our research program addresses this gap by developing new adaptive sampling methods, in which each successive stimulus depends on the subject's response to the previous stimulus. Such processes allow us to sample from the complex and high-dimensional joint distribution associated with internal representations and obtain high resolution maps of perceptual spaces.
Related publications
Another related area of interest is comparing human and machine representations. With the rise of Large Language Models and foundational multimodal machine learning, it is crucial to understand how machine learning models align with human cognition. This is essential not only for improving machine learning models but also for enhancing their interpretability and safety.
Related publications
3. Understanding cross-cultural variation in music
Music is found in every known human society, yet it is remarkably diverse across cultures. Traditional music psychology has relied predominantly on participants from a small number of countries and on Western musical materials. This sampling bias limits our ability to distinguish universal aspects of human perception from those shaped by cultural experience: similarities observed among participants may reflect shared biological mechanisms, but they may also arise from similar patterns of musical exposure.
To address this limitation, we combine computational methods with field research involving diverse populations around the world. We also analyze large-scale global cultural datasets and develop new experimental infrastructures that enable online studies with participants across many countries and cultural contexts. Together, these approaches allow us to investigate how biological, cognitive, and cultural processes interact to produce both commonality and variation in human musical perception. The same methodological framework can be extended beyond music to investigate cross-cultural commonalities and variation in other perceptual domains, including color perception and visual aesthetics.
Related publications
