Modeling Theory of Mind for Autonomous Agents with Probabilistic Programs

  • 2018-12-04 18:19:34
  • Iris Rubi Seaman, Jan-Willem van de Meent, David Wingate
  • 16

Abstract

As autonomous agents become more ubiquitous, they will eventually have toreason about the mental state of other agents, including those agents' beliefs,desires and goals - so-called theory of mind reasoning. We introduce acollection of increasingly complex theory of mind models of a "chaser" pursuinga "runner", known as the Chaser-Runner model. We show that our implementationis a relatively straightforward theory of mind model that can capture a varietyof rich behaviors, which in turn, increase runner detection rates relative tobasic (non-theory-of-mind) models. In addition, our paper demonstrates that (1)using a planning-as-inference formulation based on nested importance samplingresults in agents simultaneously reasoning about other agents' plans andcrafting counter-plans, (2) probabilistic programming is a natural way todescribe models in which each uses complex primitives such as path planners tomake decisions, and (3) allocating additional computation to perform nestedreasoning about agents result in lower-variance estimates of expected utility.

 

Quick Read (beta)

loading the full paper ...