Professor of Philosophy and Psychology and Department Head, Department of Philosophy
My research largely falls at the intersection of philosophy, cognitive science, and machine learning, using ideas and frameworks from each to inform the others. My primary research in recent years has been in computational cognitive science: developing fully-specified computational models to describe, predict, and most importantly, explain human behavior. I have focused in particular on the representations and processes that underlie complex human cognition, including causal learning/inference, concept acquisition/application, and decision-making. The cognitive architecture resulting from that research is laid out in my book, Unifying the Mind: Cognitive Representations as Graphical Models. Most recently, I have begun to explore how this high-level architecture could be implemented or instantiated in psychologically plausible lower-level models.
I have also worked on a range of problems in machine learning, principally on automated causal inference: that is, what can be learned about the causal structure of the world from different types of observational and/or experimental data? In general, I have focused on learning from unusual data sources, such as multiple distinct datasets that share only some-but-not-all variables. My current machine learning research centers on causal learning from complex time series data (e.g., with significant undersampling, or with unobserved, causally relevant variables), with a particular eye towards implications for neuroimaging and our ability to learn about the mind/brain.
I have most recently begun a research program focusing on the role of human agents in cyber-systems, principally exploring various ethical and cognitive puzzles that arise during cyber-conflicts. Most discussions of cyber-warfare, cyber-espionage, and other adversarial interactions in the cyber-domain have focused on the machines; my research centers instead on the humans who program, operate, and are affected by those machines. In particular, the ethical and strategic landscape changes significantly when we include the cognitive processes, constraints, and biases of those human agents.