How Can Memory-Augmented Neural Networks Pass a False-Belief Task?

Erin Grant, Aida Nematzadeh, Thomas L. Griffiths

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Scopus citations

Abstract

A question-answering system needs to be able to reason about unobserved causes in order to answer questions of the sort that people face in everyday conversations. Recent neural network models that incorporate explicit memory and attention mechanisms have taken steps towards this capability. However, these models have not been tested in scenarios for which reasoning about the unobservable mental states of other agents is necessary to answer a question. We propose a new set of tasks inspired by the well-known false-belief test to examine how a recent question-answering model performs in situations that require reasoning about latent mental states. We find that the model is only successful when the training and test data bear substantial similarity, as it memorizes how to answer specific questions and cannot reason about the causal relationship between actions and latent mental states. We introduce an extension to the model that explicitly simulates the mental representations of different participants in a reasoning task, and show that this capacity increases the model's performance on our theory of mind test.

Original languageEnglish (US)
Title of host publicationCogSci 2017 - Proceedings of the 39th Annual Meeting of the Cognitive Science Society
Subtitle of host publicationComputational Foundations of Cognition
PublisherThe Cognitive Science Society
Pages427-432
Number of pages6
ISBN (Electronic)9780991196760
StatePublished - 2017
Externally publishedYes
Event39th Annual Meeting of the Cognitive Science Society: Computational Foundations of Cognition, CogSci 2017 - London, United Kingdom
Duration: Jul 26 2017Jul 29 2017

Publication series

NameCogSci 2017 - Proceedings of the 39th Annual Meeting of the Cognitive Science Society: Computational Foundations of Cognition

Conference

Conference39th Annual Meeting of the Cognitive Science Society: Computational Foundations of Cognition, CogSci 2017
Country/TerritoryUnited Kingdom
CityLondon
Period7/26/177/29/17

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Computer Science Applications
  • Human-Computer Interaction
  • Cognitive Neuroscience

Keywords

  • false-belief test
  • language understanding
  • question answering
  • theory of mind

Fingerprint

Dive into the research topics of 'How Can Memory-Augmented Neural Networks Pass a False-Belief Task?'. Together they form a unique fingerprint.

Cite this