Eric Schulz

Learning to act by integrating mental simulations and physical experiments

The paper shows people balance mental simulations and physical experiments to learn action effects, adjusting effort based on trial characteristics.

Adaptive planning in human search

The paper shows humans plan ahead by searching at least 3 steps in a foraging task, using a heuristic called root-sampling that outperforms more complex models.

Active Function Learning

This study investigates how people actively search for information to learn linear functions, finding that a rule-based model best explains their behavior.

Learning list concepts through program induction

The paper introduces a concept learning paradigm using term rewriting systems to model human learning of complex conceptual relationships.

Structured priors in human forecasting

This study shows that human forecasts are guided by structured priors derived from real-world experience, leading to context-dependent predictions that align with real-world structures.

Active learning reveals underlying decision strategies

This study uses model-based active learning to determine if people use heuristics or full-information strategies in decision-making, finding they follow a weight-based strategy.

A tutorial on Gaussian process regression: Modelling, exploring, and exploiting functions

This tutorial introduces Gaussian process regression as a tool for modeling and exploring unknown functions, with examples and applications in regression, kernel assumptions, and exploration-exploitation scenarios.

Amortized Hypothesis Generation

The paper demonstrates that people use amortized inference strategies when answering related queries, caching summary statistics rather than raw samples.

Strategic exploration in human adaptive control

This study investigates how humans balance exploration and exploitation in uncertain environments, finding strategic yet conservative exploration behaviors.

Mapping the unknown: The spatially correlated multi-armed bandit

This study introduces a spatially correlated multi-armed bandit task to model how people form beliefs about unexplored payoffs and guide search behavior based on spatial correlations.