A dependence detection heuristic in causal induction to handle nonbinary variables

Abstract How humans estimate causality is one of the central questions of cognitive science, and many studies have attempted to model this estimation mechanism. Previous studies indicate that the pARIs model is the most descriptive of human causality estimation among 42 normative and descriptive mod...

Full description

Saved in:
Bibliographic Details
Main Authors: Kohki Higuchi, Tomohiro Shirakawa, Hiroto Ichino, Tatsuji Takahashi
Format: Article
Language:English
Published: Nature Portfolio 2025-04-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-91051-7
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849765474430091264
author Kohki Higuchi
Tomohiro Shirakawa
Hiroto Ichino
Tatsuji Takahashi
author_facet Kohki Higuchi
Tomohiro Shirakawa
Hiroto Ichino
Tatsuji Takahashi
author_sort Kohki Higuchi
collection DOAJ
description Abstract How humans estimate causality is one of the central questions of cognitive science, and many studies have attempted to model this estimation mechanism. Previous studies indicate that the pARIs model is the most descriptive of human causality estimation among 42 normative and descriptive models defined using two binary variables. In this study, we build on previous research and attempt to develop a new descriptive model of human causal induction with multi-valued variables. We extend the applicability of pARIs to non-binary variables and verify the descriptive performance of our model, pARIsmean, by conducting a causal induction experiment involving human participants. We also conduct computer simulations to analyse the properties of our model (and, indirectly, some tendencies in human causal induction). The experimental results showed that the correlation coefficient between the human response values and the values of our model was r = .976, indicating that our model is a valid descriptive model of human causal induction with multi-valued variables. The simulation results also indicated that our model is a good estimator of population mutual information when only a small amount of observational data is available and when the probabilities of cause and effect are nearly equal and both probabilities are small.
format Article
id doaj-art-25933baf2c7a4ec5a89ea96a3c5c6164
institution DOAJ
issn 2045-2322
language English
publishDate 2025-04-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-25933baf2c7a4ec5a89ea96a3c5c61642025-08-20T03:04:50ZengNature PortfolioScientific Reports2045-23222025-04-0115111410.1038/s41598-025-91051-7A dependence detection heuristic in causal induction to handle nonbinary variablesKohki Higuchi0Tomohiro Shirakawa1Hiroto Ichino2Tatsuji Takahashi3Chubu UniversityChubu UniversityTokyo Denki UniversityTokyo Denki UniversityAbstract How humans estimate causality is one of the central questions of cognitive science, and many studies have attempted to model this estimation mechanism. Previous studies indicate that the pARIs model is the most descriptive of human causality estimation among 42 normative and descriptive models defined using two binary variables. In this study, we build on previous research and attempt to develop a new descriptive model of human causal induction with multi-valued variables. We extend the applicability of pARIs to non-binary variables and verify the descriptive performance of our model, pARIsmean, by conducting a causal induction experiment involving human participants. We also conduct computer simulations to analyse the properties of our model (and, indirectly, some tendencies in human causal induction). The experimental results showed that the correlation coefficient between the human response values and the values of our model was r = .976, indicating that our model is a valid descriptive model of human causal induction with multi-valued variables. The simulation results also indicated that our model is a good estimator of population mutual information when only a small amount of observational data is available and when the probabilities of cause and effect are nearly equal and both probabilities are small.https://doi.org/10.1038/s41598-025-91051-7Causal inferenceDescriptive modelRational analysisNon-linear causalityCategorical data analysisJaccard index
spellingShingle Kohki Higuchi
Tomohiro Shirakawa
Hiroto Ichino
Tatsuji Takahashi
A dependence detection heuristic in causal induction to handle nonbinary variables
Scientific Reports
Causal inference
Descriptive model
Rational analysis
Non-linear causality
Categorical data analysis
Jaccard index
title A dependence detection heuristic in causal induction to handle nonbinary variables
title_full A dependence detection heuristic in causal induction to handle nonbinary variables
title_fullStr A dependence detection heuristic in causal induction to handle nonbinary variables
title_full_unstemmed A dependence detection heuristic in causal induction to handle nonbinary variables
title_short A dependence detection heuristic in causal induction to handle nonbinary variables
title_sort dependence detection heuristic in causal induction to handle nonbinary variables
topic Causal inference
Descriptive model
Rational analysis
Non-linear causality
Categorical data analysis
Jaccard index
url https://doi.org/10.1038/s41598-025-91051-7
work_keys_str_mv AT kohkihiguchi adependencedetectionheuristicincausalinductiontohandlenonbinaryvariables
AT tomohiroshirakawa adependencedetectionheuristicincausalinductiontohandlenonbinaryvariables
AT hirotoichino adependencedetectionheuristicincausalinductiontohandlenonbinaryvariables
AT tatsujitakahashi adependencedetectionheuristicincausalinductiontohandlenonbinaryvariables
AT kohkihiguchi dependencedetectionheuristicincausalinductiontohandlenonbinaryvariables
AT tomohiroshirakawa dependencedetectionheuristicincausalinductiontohandlenonbinaryvariables
AT hirotoichino dependencedetectionheuristicincausalinductiontohandlenonbinaryvariables
AT tatsujitakahashi dependencedetectionheuristicincausalinductiontohandlenonbinaryvariables