Empirical Evaluation of Observational Causal Induction Models

DOI

Bibliographic Information

Other Title
  • 観察的因果推論モデルの実験的評価

Abstract

<p>Humans learn,reason,and utilize causal relationships - the connection between one event and the occurrence of another - to diagnose causes,explain reasons,and make predictions about the future in order to adapt to uncertain environments.Causal inference is closely related to the field of artificial intelligence,where quick learning and decision-making from small amounts of data,like humans,are supported by efficient causal reasoning.Based on this background,the purpose of this paper is to elucidate how humans learn and judge causal relationships from statistical information.This paper focuses specifically on observational causal inference and proceeds by referring to causal induction models proposed so far.We conduct multiple experiments and analyses,including a direct comparison of the descriptive performance of the Dual-Factor Heuristic (DFH) proposed by Hattori and Oaksford (2007) and the proportion of Assumed-to-be Rare Instances (pARIs) proposed by Takahashi et al.(2011)showing that pARIs are more suitable for human judgment</p>

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top