The general focus of Mark Steyvers, Ph.D.'s research in cognitive science is high-level cognition. The goal is to understand how people self-reflect, learn, and make decisions through empirical studies and computational modeling. He has used these insights on human intelligence to contribute not only to cognitive science and psychology, but also advance machine learning research.
Recently, in projects on human-AI (artificial intelligence) collaboration, Steyvers has started to investigate how humans can collaborate with AI / Machine learning models to amplify and augment human decision-making. In this human-AI research, he studies human capacities such as theory-of-mind (how do people build mental models of each other?) and metacognition (how do people assess their own knowledge limitations?) to create new goals and benchmarks for AI/ML models — how can an AI learn to help a person given what the AI knows about the person’s limitations as well as its own? In another research thread, he is interested in understanding how people acquire skills across multiple cognitive tasks (memory, attention, reasoning) and explain individual differences across the lifespan. Steyvers tests these ideas on large-scale datasets from cognitive training platforms.