Chemla, Emmanuel, Paul Egré, and Philippe Schlenker. (2012). “Predicting moral judgments from causal judgments”.

Manuscript LSCP and Institut Jean-Nicod, Paris, and NYU.

[Full paper in pdf]

[BibTeX reference]

Note: This work is not a study in mainstream semantics, but it shows how to use semantic techniques to model and test non-purely linguistic phenomena.

Abstract: Several factors have been put forward to explain the variability of moral judgments for superficially analogous moral dilemmas, in particular in the paradigm of trolley cases. In this paper we elaborate on Mikhail's view that (i) causal analysis is at the core of moral judgments and that (ii) causal judgments can be quantified by linguistic methods. According to this model, our moral judgments depend both on utilitarian considerations (whether positive effects outweigh negative effects) and on a representation of the causal structure of the action (whether the negative effects are essentially side-effects rather than main goals). However the exact contribution of each factor, as well as the precise way in which causal considerations interact with utilitarian considerations, has yet to be quantified and investigated. We present several variations on trolley dilemmas in which subjects had to assess the morality of the action and to evaluate their preference between two competing descriptions of the scene ('a caused the death of m, thereby saving n' vs. 'a saved n, thereby causing the death of m'). Our main finding is that moral judgments are highly correlated with causal judgments in terms of such descriptions, allowing us to predict the former from the latter. Furthermore, we observe that the effect of causal judgments on the relative permissibility of actions is felt even in anti-utilitarian scenarios, namely scenarios for which the proposed action diminishes aggregate utility.