MADLAB

Research Areas

• Hybrid human-machine algorithm systems
  • How can we leverage the collective intelligence of human-AI systems? What are the best ways for humans and algorithms to work together? How can an AI algorithm assess the capabilities of a human, and how can humans learn to trust the advice of the algorithm? How do we develop computational models of trust and understand the mental model the human user develops of the AI algorithm?
• Bayesian computational modeling
  • Throughout most computational modeling projects in the lab, the goal is to use Bayesian methods to infer model parameters and assess uncertainty. When applied to large-scale data sets, how do we model the joint performance of individuals across cognitive tasks? How can we leverage Bayesian methods to understand what individuals have learned and what they are capable of learning?
• Learning and Memory
  • How can we leverage large-scale data to analyze the learning trajectories across individuals and cognitive tasks? How do we develop computational models to explain what is learned when individuals improve a skill?
• Cognitive Skill Acquisition and Transfer
  • How can we leverage large-scale data from cognitive training platforms to analyze transfer? How can we build computational models to understand the effects of transfer?
• Metacognition
  • How do we know what we know (or do not know)? Can we build computational models for metacognition that can monitor and self-direct learning? What are the limits of human metacognitive abilities? What theory of metacognition would explain the nature of these limits?
• Machine learning
  • How can we use insights from cognitive science to build better machine-learning models? For example, how do we endow artificial learning algorithms with the capability to learn new tasks as quickly as humans? How can artificial learning algorithms understand task instructions?

Current Lab Members

Former Students

 

---