CompCog: Developing a Theory of Causal Learning over Time

从经验中学习是人类如何成功预测和适应不断变化的世界的关键工具，无论是教师试图找出哪种教学技术对课堂最有效，医生试图找出哪种药物对病人最有效，还是商人试图找出哪种营销策略能产生最大的销售额。这个项目的目标是更好地理解人们如何在一个高度复杂的世界中学习因果关系并做出因果判断。这将有助于预测人类何时可能做出正确的决定，何时可能做出错误的决定，这对个人和社会来说都可能是昂贵的。关于人们如何学习因果关系的研究主要集中在独立的、相同分布的、横断面的情况下的学习。然而，人类也学习时间分布的事件之间的因果关系，这往往涉及非平稳和自相关的信息。这项工作的第一个目标是确定人们在这些纵向情况下用来学习因果关系的过程。第二个目标是确定人们在纵向和横向环境中采用不同学习策略的能力。结果还将用于开发在线模拟，以帮助学生学习如何在设计和分析研究时做出良好的因果判断。

Learning from experience is a critical tool for how humans successfully predict and adjust to their changing world, whether a teacher trying to figure out which pedagogical technique will work best for a class, a doctor trying to figure out which medicine will work best for patient, or a business person trying to figure out which marketing strategy will produce the largest sales. The goal of this project is to better understand how people learn cause-effect relations and make causal judgments in a highly complex world. This will facilitate predicting when humans are likely to make good decisions and when humans are likely to make bad decisions, which can be costly for the individual and for society. The research on how people learn causal relations has primarily focused on learning in independent and identically distributed, cross-sectional situations. However, humans also learn causal relations among events that are distributed in time, which often involves non-stationary and autocorrelated information. The first goal of this work is to identify processes people use to learn causal relationships in these longitudinal situations. The second goal is to identify people's ability to employ different learning strategies adaptively for both longitudinal and cross-sectional environments. The results will also be used to develop online simulations to help students learn how to make good causal judgments when designing and analyzing research.