UC Berkeley

The explosion of data generated during people's interactions online presents an opportunity for cognitive scientists to evaluate their models on real-world tasks outside the constraints of the psychology laboratory. In addition, new internet-based experimental paradigms and crowd-sourcing tools have reduced the challenge of data collection, offering researchers the opportunity to apply powerful new data-driven modeling techniques in domains which were previously limited to hand-tuned theoretic models. In this thesis I illustrate the opportunity internet-scale data offer cognitive scientists using three in-depth case-studies. Each study focuses on a different fundamental question in cognitive science -- search in associative memory, generalization, and risky-choice behavior, respectively -- and demonstrates how larger datasets and new online experimental methods make analyses which would otherwise prove difficult or impossible using traditional in-lab techniques feasible.