Learning, Awareness, Optimism, and Confidence

Abstract

Major Goals: The goal of this project is to investigate operations on causal models that will be critical making best use of causal information. The initial focus will be on two operations: combining models and abstracting models. With regard to combining models, the assumption is that experts will provide causal models to a policymaker, who wants to combine (the information in) these models to reach her decision. Bradley, Dietrich, and List have provided arguably reasonable desiderata for combining causal models, and showed that there is no way of combining causal models so as to satisfy all their desiderata. The hope is that we can provide conditions under which models are compatible. The general approach to combining experts judgements would then be to combine the experts models when they are compatible, and if not, just place a probability on each model being the right model, using relatively standard techniques based on the perceived reliability of the experts who proposed them. Intuitively, if the experts models are not compatible, then the experts are disagreeing, so we should not try to combine their models; rather, we should just assign a likelihood to each model being right. On the other hand, if the models are compatible, then we should consider a model that takes into account both experts information. With regard to abstraction, suppose that an expert provides a detailed causal model, involving many variables. Such a detailed model may be far too complicated for a policymaker to understand and work with. What the policymaker wants is a high-level macro model of the situation. Such a high-level model can result from combining many variables at the micro level into one macro variable. There have been recent arguments made that such a high-level model actually can, in a precise sense, provide more information than a detailed model. A high-level model can also be helpful when it comes to deciding appropriate interventions to perform.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 15, 2018
Accession Number
AD1070285

Entities

People

  • Joseph Halpern

Organizations

  • Cornell University

Tags

Communities of Interest

  • Autonomy

DTIC Thesaurus Topics

  • Abstracts
  • Algorithms
  • Artificial Intelligence
  • Automata
  • Autonomous Agents
  • Computations
  • Decision Theory
  • Distributed Computing
  • Electronic Commerce
  • Game Theory
  • Information Operations
  • Learning
  • Multiagent Systems
  • Probability
  • Standards
  • Students
  • Technology Transfer

Readers

  • Artificial Intelligence
  • Computational Modeling and Simulation