Item

Abstract

In the present paper we show that a computational model of affective moral decision making can fit human behavior data obtained from an empirical study on criminal decision making. By applying parameter tuning techniques on data from an initial sample, optimal fits of the affective moral decision making model were found supporting the influences of honesty/humility, perceived risk and negative state affect on criminal choice. Using the parameter settings from the initial sample, we were able to predict criminal choices of participants in the holdout sample. The prediction errors of the full model were fairly low. Moreover, they compared favorably to the prediction errors produced by constrained variants of the model where either the moral, rational or affective influences or a combination of these had been removed.