Running head: THE RELEVANCE OF ALTERNATIVES Unifying morality’s influence on non-moral judgments: The relevance of alternative possibilities

Past work has demonstrated that people’s moral judgments can influence their judgments in a number of domains that might seem to involve straightforward matters of fact, including judgments about freedom, causation, the doing/allowing distinction, and intentional action. The present studies explore whether the effect of morality in these four domains can be explained by changes in the relevance of alternative possibilities. More precisely, we propose that moral judgment influences the degree to which people regard certain alternative possibilities as relevant, which in turn impacts intuitions about freedom, causation, doing/allowing, and intentional action. Employing the stimuli used in previous research, Studies 1a, 2a, 3a, and 4a show that the relevance of alternatives is influenced by moral judgments and mediates the impact of morality on non-moral judgments. Studies 1b, 2b, 3b, and 4b then provide direct empirical evidence for the link between the relevance of alternatives and judgments in these four domains by manipulating (rather than measuring) the relevance of alternative possibilities. Lastly, Study 5 demonstrates that the critical mechanism is not whether alternative possibilities are considered, but whether they are regarded as relevant. These studies support a unified framework for understanding the impact of morality across these very different kinds of judgments. THE RELEVANCE OF ALTERNATIVES 3 Unifying morality’s influence on non-moral judgments: The relevance of alternative possibilities A series of recent studies have shown that people’s moral judgments can impact their intuitions about issues that might appear to be straightforward matters of fact. This effect was noted early on for intuitions about whether an agent acted intentionally (Knobe, 2003), but it soon became clear that a very similar pattern could be found in numerous other domains. Among other things, people’s moral judgments also influence their intuitions about whether an agent acted freely, whether an action caused some further outcome, and whether an agent did something or merely allowed it to happen (Cushman, Knobe & Sinnott-Armstrong, 2008; Hitchcock & Knobe, 2009; Phillips & Knobe 2009; Young & Phillips, 2011). One way to explain these phenomena would be to offer an independent account for each of the separate effects. So one could construct a hypothesis about why moral considerations influence intuitions about intentional action (Guglielmo & Malle, 2010; Machery, 2008; Nichols & Ulatowski, 2007; Uttich & Lombrozo, 2010), then another separate hypothesis about why moral considerations impact intuitions about causation (Menzies, 2010; Sytsma, Livengood, & Rose, 2012), and so on. (Some have even argued explicitly that there cannot be a unified explanation of all of these effects; Hindriks, 2014.) Here, we pursue the exact opposite approach. Rather than looking separately at each individual effect, we want to consider the widespread influence of moral cognition in these different domains, and to offer a unified explanation that applies to them all. Specifically, we will suggest that all these effects can be explained in terms of a very general principle involving the relationship between people’s moral judgments and their way of understanding alternative possibilities. The Role of Possibilities Research in a number of different fields has independently argued for the importance of THE RELEVANCE OF ALTERNATIVES 4 alternative possibilities, including work in philosophy (e.g., Bennett, 2003; Lewis, 1973), linguistics (Kratzer, 2012; for a review, see Portner, 2009), psychology (Kahneman & Miller, 1986; Byrne, 2005, for a review see, Roese, 1997), and computer science (e.g., Bello, in press; Halpern & Pearl, 2005; Pearl, 2000). In each of these cases, the central insight has been that our understanding of the things that occur is shaped in some fundamental way by our understanding of alternative possibilities that could have occurred but actually did not. Intriguingly, existing theoretical work has already suggested that alternative possibilities play a role in each of the domains in which we find these surprising effects of moral judgment. As one example, consider judgments about freedom. Within existing theoretical work, it is widely acknowledged that the question as to whether someone acted freely depends, in part, on whether it would have been possible for that person to have done something else instead (Aquinas, 1273/1920; Berlin, 1969/2002; Aristotle, 350 BC/2002). Thus, on one view, claims about freedom like (1a) depend on claims about alternative possibilities like (1b). (1) a. She did x freely. b. If she had wanted not to do x, she would not have done x. Similarly, within existing work on judgments about causation, it is widely believed that causal judgments in some way rely on representations of alternative possibilities (Lewis, 1973; Pearl, 2000). In particular, a number of accounts of causation suggest that causal claims like (2a) depend at least partially on claims about alternative possibilities like (2b). (2) a. Event x caused event y. b. If x had not happened, y would not have happened. Within existing work in this tradition, there is a great deal of controversy about precisely how to spell out the relationship between alternative possibilities and judgments about freedom, causation, etc. (Halpern & Hitchcock, 2014; Nduibuisi & Byrne, 2013; Woodward, 2004). However, the details THE RELEVANCE OF ALTERNATIVES 5 of this controversy will not concern us here. The key point is simply that judgments in each of these domains depend in some way on how we understand the alternative possibilities that could have happened, but didn’t. Across a number of fields, existing work on alternative possibilities has also introduced a further idea that plays a central role in the present hypothesis: People do not treat all alternative possibilities equally. Instead, they regard certain possibilities as relevant, while treating others as completely irrelevant (Halpern & Hitchcock, 2014; Lewis, 1973; Portner, 2009; Roese, 1997). To the extent that they regard an alternative possibility as irrelevant, they will tend to ignore it entirely, and it will play little role in their judgments in any domain. To give one simple example, suppose that a group of people is giving a presentation, and they end up doing a terrible job. As we consider what actually occurred in this case, we might regard certain alternative possibilities as especially relevant. For example, possibilities in which they had spent more time preparing might seem especially relevant. Or perhaps even possibilities in which they had simply decided not to give the presentation at all. But then there are numerous other possibilities that we would regard as completely irrelevant. Thus, we would never regard as relevant the possibility that the presentation could have been interrupted by a freak tornado, or the possibility that the earth’s gravitational field could have suddenly ceased to exist. People might be capable of entertaining possibilities like these if forced to, but all the same, they would regard them utterly pointless, not even worthy of the slightest consideration. Most importantly for present purposes, the distinction between relevant and irrelevant possibilities has been invoked to explain judgments in the specific domains under consideration here. Take the example of causation. As we noted above, it seems that people only regard a factor as causal if, when considering alternative possibilities in which that factor does not occur, the outcome also does not occur (Lewis, 1973; Pearl, 2000). But theoretical work suggests that it is not enough THE RELEVANCE OF ALTERNATIVES 6 for this to be true; possibilities in which this factor does not occur must also be relevant. In cases where such possibilities are regarded as irrelevant, people will simply reject the corresponding causal claim (e.g., Halpern & Hitchcock, 2014; Schaffer & Blanchard, in press). To illustrate this point, consider the causal claim (3a) and the corresponding claim about an alternative possibility (3b). (3) a. The earth’s gravitational field caused that disastrous presentation. b. If the earth’s gravitational field had not been present, the disastrous presentation would not have occurred. The claim (3b) is surely true. Yet at the same time, possibilities in which the earth’s gravitational field is not present seem entirely irrelevant, and thus the theory predicts that the corresponding causal claim should also be seen as absurd. As this example helps to make clear, it is critical to distinguish between the degree to which people actively consider an alternative possibility and the degree to which they regard that alternative possibility as relevant. Previous work in social psychology has explored the impact of counterfactual reasoning, and this work has demonstrated that actively considering a particular possibility can have an important impact on many aspects of human behavior (for reviews, see Byrne, 2005; Epstude & Roese, 2008; Roese, 1997). The focus of the present paper, however is on a somewhat different phenomenon. Independent of the question of whether or not people consider a particular possibility, there seems to be a question as to whether people regard that possibility as relevant. (Even if people are specifically instructed to reason about the possibility that the earth’s gravitational field could suddenly disappear, they might continue to regard this possibility as completely irrelevant.) As far as we know, this approach has not yet been directly tested in experimental work, but it has been developed in considerable technical detail within the existing theoretical literature in several different fields (e.g., Bello, in press; Halpern & Hitchcock, 2014; Knobe, 2010; Kratzer, 2012; Schaffer & Blanchard, in press). THE RELEVANCE OF ALTERNATIVES 7 If we are seeking a unified way of explaining why people’s moral judgments influence their judgments in the domains of freedom, cause, etc., then the relevance of alternative possibilities may help. That is, if we can show that people’s moral judgments influence their judgments of the relevance of alternative possibilities, then we may be able to provide a unified account of morality’s influence in all four of these areas. We propose to test this basic model (Fig. 1) in the current paper. Figure 1. Proposed model with the relevance of alternative possibilities mediating the influence of morality in the domains of freedom, causation, doing vs. allowing, and intentional action. Morality and the Relevance of Alternative Possibilities Thus far, we have been considering the evidence that judgments of the relevance of alternative possibilities influence intuitions in each of the four domains. We now turn to the other link in our model – the claim that moral judgments influence judgments about the relevance of alternative possibilities. Specifically, there is a general tendency to regard alternative possibilities as more relevant to the extent that they involve replacing something morally bad (in the actual world) THE RELEVANCE OF ALTERNATIVES 8 with something morally good (in the alternative possibilities). Some form of this basic idea can be found across a wide variety of different theoretical accounts that have been spelled out within quite different theoretical frameworks (e.g., Halpern & Hitchcock, 2014; Knobe & Szabo, 2013; Schaffer & Blanchard, in press). The precise technical implementation varies considerably from one framework to the next, but the core idea that is shared across all of these accounts is a highly intuitive one. Suppose you believe that a certain aspect of the way people typically treat each other is fundamentally morally wrong. You might then see it as highly relevant to consider alternative possibilities in which people treated each other in the way you believed to be morally good. If you saw a man insulting a homeless person, for example, the alternative that he could have instead tried to help the homeless person clearly seems relevant. Now, by contrast, suppose you believe that a certain aspect of the way people typically treat each other is morally good. In that latter case, you might regard it as completely irrelevant to consider alternative possibilities in which, for some reason, people treated each other in the way you believe to be morally bad. It clearly does not seem relevant to think about how the man could have insulted the homeless man instead of helping him. The Present Studies Past work offers theoretical support for the idea that the influence of morality may occur by altering people’s intuitions about the relevance of possibilities. Yet to date, there has been little empirical work that directly investigates this relationship. This is our aim in the present studies. More precisely, we propose that morality influences the degree to which people judge certain alternative possibilities as relevant, which in turn impacts people’s intuitions in non-moral domains. To test this model, we took the materials from four previously published studies that originally demonstrated morality’s influence in the domains of freedom, causation, doing/allowing, and intentional action. The use of these previous materials naturally limits researcher degrees of THE RELEVANCE OF ALTERNATIVES 9 freedom and experimenter bias (Simmons, Nelson, & Simonsohn, 2011; Strickland & Suben, 2012). Further constraining researcher degrees of freedom, we also explored each domain using exactly the same methods. Specifically, within each domain, we conducted two studies: one that checked for mediation by relevance of alternatives and one that manipulated relevance of alternatives. For each of the mediation studies, we replicated an existing experiment and then added a further measure in which participants were directly asked about the relevance of certain alternatives. We predicted that the different conditions of the original experiment would lead to different judgments on the measure of relevance of alternatives, which would in turn mediate the impact of condition on the original dependent variable. For the manipulation studies, we took the morally neutral condition of each of the original experiments and then tried to manipulate the relevance of alternatives using an entirely non-moral method. If morality affects judgments of freedom, causation, doing/allowing, and intentional action by changing the relevance of alternative possibilities, then morality should not be special in its influence on these domains. In fact, any factor that influences the relevance of alternative possibilities should also impact judgments in these various domains. Accordingly, to manipulate the extent to which participants regarded alternatives to the agents’ actions as relevant, participants were instructed to write a brief paragraph about what else the agent could have done. (Participants in a control condition were simply asked to summarize the story). Admittedly, the effect of this novel way of manipulating the relevance of alternative possibilities may not be as strong as the original effects of morality observed in the previous studies. Still, as this manipulation parallels the proposed mechanism, we expect that participants who are asked to engage in writing about alternative possibilities (vs. summarize the story) will show a pattern of judgments similar to those observed in the morally bad (vs. morally neutral) actions. These experimental methods are applied to judgments of freedom (Study 1a-b), causation THE RELEVANCE OF ALTERNATIVES 10 (Study 2a-b), doing vs. allowing (Study 3a-b), and intentional action (Study 4a-b). In each case, we find two pieces of support for the proposed model. First, we find that judgments of the relevance of alternative possibilities mediate the effect of morality on judgments in these four domains. However, the evidence from these mediation analyses is equivocal because the data fit alternative mediation models. Second, and more conclusively, we provide direct evidence of the role of alternative possibilities by non-morally manipulating the relevance of alternatives and then demonstrating that all four kinds of judgments are affected by this manipulation in precisely the same way that they were affected by morality. Having considered relevant alternatives to the agent’s action, participants judge that the actor was less forced, more of a cause, did something rather than merely allowed it to happen, and acted more intentionally. Study 5 then provides a demonstration that it is the perceived relevance of alternatives (rather than simply the process of considering of alternative possibilities) that affects participants’ judgments. Taken together, these five studies provide support for a unifying framework in which the impact of morality on a wide array of disparate judgments can be accounted for by morality’s influence on the relevance of alternative possibilities. Study 1: Freedom and Force Past research has found that morality impacts judgments about whether an agent performed an action freely or was simply forced to perform that action (Phillips & Knobe, 2009; Young & Phillips, 2011). In one study, participants were asked to read a scenario in which a ship captain saved his ship from sinking either by doing something morally bad (throwing his wife overboard) or by doing something morally neutral (throwing his wife’s cargo overboard). Participants were less likely to judge that the ship captain was forced to throw something overboard when doing so was morally bad than when it was not. In theoretical accounts of freedom (Aquinas, 1273/1920; Berlin, 1969/2002; Aristotle, 350 THE RELEVANCE OF ALTERNATIVES 11 BC/2002), whether or not an agent is rightly said to have acted freely depends critically on what other actions the agent could have done instead. It has proven remarkably difficult to say precisely how it is that the concept of freedom relates to these other possible actions, and which other possibilities bear on the question as to whether the agent acted freely (see, e.g., Cova, 2014; Cova & Kitano, 2014; Miller & Feltz, 2011; Woolfolk, Doris & Darley, 2006). However, what we rely on here is simply the idea that alternative possibilities play some essential role in people’s judgments about freedom. Thus, one possible way to explain the impact of moral judgments on judgments of freedom is to argue that participants’ moral judgments influenced the extent to which they found it relevant to consider the alternative actions that the agent could have done instead. We test this possibility using both a mediational analysis (Study 1a) and a direct manipulation of the relevance of alternative possibilities (Study 1b). Study 1a Mediation If people’s moral judgments are changing which alternative possibilities they consider to be relevant when reading the vignette, we predict that participants will judge that it is more relevant to consider the possibility that the ship captain could have not thrown his wife (vs. his wife’s cargo) overboard. Moreover, we predict that the previously observed impact of morality on judgments of freedom will be mediated by these judgments of relevance. Method Participants. Four hundred participants (148 female, 1 unreported, mean age = 32.65, SD=10.07) were recruited on Amazon’s Mechanical Turk in exchange for a small monetary payment. Procedure. All participants read the vignette about the ship captain (Phillips & Knobe, 2009). Those in the morally neutral condition read about the captain throwing his wife’s cargo THE RELEVANCE OF ALTERNATIVES 12 overboard, whereas those in the morally bad condition read about the captain throwing his wife