A Semantics for Causing, Enabling, and Preventing Verbs using Structural Causal Models

This repository contains the experiments, data, analyses, and figures for the paper "A Semantics for Causing, Enabling, and Preventing Verbs using Structural Causal Models" by Angela Cao* ([email protected]), Atticus Geiger*, Elisa Kreiss*, Thomas Icard, and Tobias Gerstenberg ([email protected]).

(* indicates equal contribution)

Abstract

When choosing how to describe what happened, we have a number of causal verbs at our disposal. In this paper, we develop a model-theoretic formal semantics for nine causal verbs that span the categories of CAUSE, ENABLE, and PREVENT. We use structural causal models (SCMs) to represent participants' mental construction of a scene when assessing the correctness of causal expressions relative to a presented context. Furthermore, SCMs enable us to model events relating both the physical world as well as agents' mental states. In experimental evaluations, we find that the proposed semantics exhibits a closer alignment with human evaluations in comparison to prior accounts of the verb families.

Demo

You can see a demo of the experiment here.

Repo structure

.
├── exp-code
│   ├── binary-code
|   ├── scale-code
│   └── videos
├── analysis
│   ├── R script
│   ├── data1
│   ├── data2
│   └── figures
└── video-generation
    ├── script
    ├── script-output
    └── images

exp-code

binary-code

• Experimental backend for our binary-choice experiment, using _magpie. It was hosted using Netlifly, with the backend hosted on Heroku.

scale-code

• Prior to running our binary-choice experiment, we also ran a similar version where participants additional had to indicate "how confident" they were in their choice. Data from our paper only reports on the binary-choice experiment, as the results from the scale version showed similar results.

videos

• GIFs used in the experiment were first pdfs generated by our video generation script, and then converted to GIF form using a GIF converter.

analysis

R script

• analysis script for reported results and corresponding pre-processing

data1

• data collected from the experiment reported in the paper (worker IDs are replaced with a unique number)

data2

• data collected from a preliminary scale version of our experiment (worker IDs are replaced with a unique number)

figures

• figures outputted from the analysis script

video generation

script

• Python script used to generate .tex for compilation into pdf versions of the video stimuli

script-output

• pdf outpts from the script

images

• copyright-free images of entities needed for the script, such as the farmer, wizard, and rock