Efficient Kill-Save Ratios Ease Up the Cognitive Demands on Counterintuitive Moral Utilitarianism Bastien Trémolière Université de Toulouse Jean-François Bonnefon Centre National de la Recherche Scientifique The dual-process model of moral judgment postulates that utilitarian responses to moral dilem- mas (e.g., accepting to kill 1 in order to save 5) are demanding of cognitive resources. Here we show that utilitarian responses can become effortless, even when they involve to kill someone, as long as the kill-save ratio is efficient (e.g., 1 is killed in order to save 500). In Experiment 1, participants responded to moral dilemmas featuring different kill-save ratios under high or low cognitive load. In Experiments 2 and 3, participants responded at their own pace or under time pressure. Efficient kill-save ratios promoted utilitarian responding, and neutered the effect of load or time pressure. We discuss whether this effect is more easily explained by a parallel- activation model or by a default-interventionist model. Keywords: Moral cognition; Utilitarianism; Time pressure; Cognitive load Is it morally permissible to kill one in order to save many? Different people react differently to this dilemma. Those who say yes are considered to give the utilitarian response, and the core claim of the dual-process approach to moral judgment (Greene, Sommerville, Nystrom, Darley, & Cohen, 2001; Greene, Nystrom, Engell, Darley, & Cohen, 2004) is that giving this response requires controlled, effortful mental processes. If the utilitarian response is controlled rather than automatic, manipulations that temporarily decrease cognitive resources (e.g., concurrent load or time pressure) should de- crease its frequency. Although Greene, Morelli, Lowenberg, Nystrom, and Cohen (2008) failed to observe such an ef- fect of cognitive load, Trémolière, De Neys, and Bonnefon (2012) did obtain the effect by using an extra-strong manipu- lation of load. In parallel, Suter and Hertwig (2011) observed that utilitarian responses were less frequent under time pres- sure, at least for a subset of their dilemmas; and Greene et al. (2008) observed that utilitarian responses took longer when participants were under cognitive load. The claim that utilitarian responses to moral dilemma re- quire controlled, effortful cognitive processing is a pillar of current research on moral thinking. Recently though, Kahane et al. (2011) argued that utilitarian responses to moral con- flicts did not necessarily imply cognitive effort. They noted that the evidence for this claim had always been obtained with dilemmas for which the utilitarian response was highly counterintuitive, as it implied to kill or inflict severe harm. They then reported behavioral and neuroimaging data sug- gesting that giving the utilitarian response became intuitive when it implied a milder transgression, such as lying to pre- vent emotional or physical pain. The conclusion of the arti- cle was that the intuitive vs. counterintuitive dichotomy was perhaps more relevant to moral reasoning than the utilitarian vs. non-utilitarian dichotomy. Our goal in this article is to nuance this conclusion, and to demonstrate that utilitarian responses can be delivered ef- fortlessly, even when they imply a strong transgression such as killing. The basic idea underlying our experiments is that the moral acceptability to kill one in order to save N lives will increase with N, to the point that utilitarianism will be- come an undemanding response for large values of N. This is what we call the kill-save ratio effect : Efficient kill-save ratios (e.g., kill 1 to save 500) would encourage utilitarian re- sponses, while making them so undemanding that they would survive manipulations of cognitive load or time pressure. The novel idea here is that highly efficient ratios might generate effortless utilitarian judgments – not that they might encourage utilitarian judgments, period. Many studies demonstrated that people are sensitive to the ratio of posi- tive vs. harmful consequences of an action, when assessing the acceptability of this action. This is usually done by ask- ing people how many saved lives (or trees, or species) would make it acceptable to sacrifice one life (or tree, or species). These studies are typically investigating the conditions un- der which harmful actions are considered as morally accept- able (e.g., Baron & Leshner, 2000; Bartels & Medin, 2007; Nichols & Mallon, 2006; Ritov & Baron, 1999). We expect that better kill-save ratios will increase the moral acceptabil- ity of sacrificing one to save many, and thus promote utili- tarian thinking, but we are chiefly interested in the cognitive cost of utilitarian thinking as a function of kill-save ratio. Interestingly, the 12 scenarios used in Greene et al. (2008) featured vastly different kill-save ratios, and we were able to re-analyze these data. The three scenarios with the most inefficient kill-save ratios (1:1 to 1:5) elicited about 44% util- itarian responses, whereas the three scenarios with the most