On the normative significance of experimental moral psychology

Abstract

Abstract Experimental research in moral psychology can be used to generate debunking arguments in ethics. Specifically, research can indicate that we draw a moral distinction on the basis of a morally irrelevant difference. We develop this naturalistic approach by examining a recent debate between Joshua Greene and Selim Berker. We argue that Greene's research, if accurate, undermines attempts to reconcile opposing judgments about trolley cases, but that his attempt to debunk deontology fails. We then draw some general lessons about the possibility of empirical debunking arguments in ethics. Keywords: Consistency ReasoningDebunkingJoshua GreeneMoral Psychology Acknowledgements For helpful feedback we are grateful to Selim Berker, Ronald de Sousa, Brian Fiala, Joshua Greene, Thomas Hurka, Shaun Nichols, and Mark Timmons. Earlier drafts of the essay were presented at the 2011 CPA meeting in Fredericton, NB and the 2011 SPP meeting in Montreal, QC. Victor Kumar's work was supported by a fellowship from the Social Sciences and Humanities Research Council of Canada. Notes Notes [1] Berker (Citation2009, pp. 325–329) argues, more specifically, that neuroscientific research, including Greene's, is not directly relevant to normative ethics. As will become clear, neuroscientific research is inessential to Greene's main argument, as Greene (Citation2010, pp. 14–15) says, but contributes to crucial supporting arguments (see section 5). [2] Many philosophers will want to take issue with Greene's classification of moral judgments as characteristically deontological and characteristically consequentialist. We are sympathetic to this worry. Although we are open to Greene's idea of defining "deontological thinking" and "consequentialist thinking" by appeal to paradigm cases of the judgments associated with each (alleged) mode of moral thinking, we doubt that any such attempt will produce a classification that fits Greene's aims. (Thanks to Selim Berker for discussion.) Greene's argument fails, we will show, even if his classification of judgments about Footbridge and Bystander is granted for the sake of argument. [3] In fact, Greene et al.'s (Citation2009) most recent study on Footbridge, Bystander, and other trolley cases indicates that the distinction embodied in the principle of double effect between intended harms and unintended but merely foreseen harms (or something like this distinction) also accounts for the variation among subjects' responses to Footbridge and Bystander. More on this in section 6. [4] A rather significant assumption in experimentally based debunking arguments is that philosophers' intuitions have the same bases as the intuitions of psychological subjects. Otherwise it would not follow that philosophers' intuitions in particular are called into question. We won't discuss this here, except to note that it merits attention. [5] Berker and Greene express the conclusion of the argument from morally irrelevant factors differently. Berker says, "so, deontological intuitions, unlike consequentialist intuitions, do not have any genuine normative force" (2009, p. 321). Greene concludes that deontological judgments about Footbridge are "unreliable" (Citation2010, p. 16). These are explanations for what we take to be the main issue in a debunking argument: that the relevant judgments are epistemically challenged and so one must withhold judgment. Thus, in our formulation of the conclusion, we say that the deontological judgments are unwarranted. [6] In unguarded moments, Greene misrepresents his argument as giving decisive reason to reject deontology in favor of consequentialism. Evaluation of trolley cases, a single class of test cases for deontology and consequentialism, does no such thing. The matter is of course more complicated than this and many other considerations bear on the relative evaluation of consequentialism and deontology, as Greene is aware. His more grand ambition is to debunk all of the intuitions upon which he thinks deontology rests (Greene, Citation2010, p. 21). [7] Greene does not offer arguments for the moral-epistemological assumption per se. He argues not for the philosophical claim that the principle evidence for deontology is intuitions, but for the psychological claim that, in general, deontological commitments are based on unconsciously generated, emotion driven intuitions (Greene, Citation2008). Drawing on Baron (Citation1994), Haidt (Citation2001), and his own research, Greene argues that the arguments and principles people offer in support of their deontological beliefs are post hoc rationalizations and therefore causally downstream from those beliefs. So, whereas we have insisted on the philosophical claim that deontology is justified principally (if at all) by intuitions, Greene argues instead for the psychological claim that deontologists base their beliefs principally on intuitions. In our view, the philosophical claim is needed to rebut the charge of "genetic fallacy." For someone might argue that although intuitions are the initial cause of deontological moral theorizing, more venerable reasoning later sustains and justifies deontology. [8] About the final possibility, consequentialists might put the point more finely: that integrity, properly conceived, is compatible with maximizing the good. [9] Berker's "most pressing worry" (Citation2009, p. 325) in his essay is that neuroscience does no work in Greene's argument. Greene, as we explain in section 5, appeals to neuroscientific findings to argue that characteristically deontological judgments are generated by emotional processing in system 1, while characteristically consequentialist judgments are generated by cognitive processing in system 2, Berker argues that this hypothesis about the neural mechanisms underlying moral evaluation is not germane to the argument from morally irrelevant factors, as these systems are defined in section 5 below. Whether Greene has impugned the epistemic status of our deontological intuitions, Berker says, is "purely a function of what sorts of features out there in the world they are each responding to" (2009, p. 325) and not whether the intuitions derive from emotional systems in the brain. Greene says, and in our view he is right, that "this shouldn't be Berker's most pressing worry" (2010, p. 14). By itself, Berker's objection claims only that the empirical premise in the argument from morally irrelevant factors is supported by Greene's behavioral findings but not by his neuroscientific findings. Greene can simply concede, as he does, that neuroscience is "not essential to [his main] normative argument" (2010, p. 14). Greene does insist that his neuroscientific research, though inessential, supports his dual process model, and so is indirectly relevant to the empirical premise in his argument. In any case, Berker's main worry does not even purport to challenge the soundness of Greene's argument from morally irrelevant factors. If Greene's argument against deontology fails, it cannot be simply because neuroscience is irrelevant to that argument. However, another interpretation of Berker's main worry is available in light of the fact that there are two versions of Greene's argument from morally irrelevant factors. Our focus has been on the most recent incarnation: the judgment about Footbridge is based on personal harm, a morally irrelevant factor, and is therefore unwarranted. Greene's original argument, according to Berker, is broader: deontological judgments in general are based on personal harm, a morally irrelevant factor, and are therefore unwarranted. Berker argues that experimental moral psychology (he says "neuroscience") does no work in the argument because Greene provides no way of defining deontological and consequentialist judgments independently of what he thinks they are responses to. So, Berker thinks, the claim that deontological judgments are responses to personal harm is a question-begging armchair stipulation and not an empirical claim—not supported by any of Greene's research. In our view, this objection fails too. Greene's approach is to identify paradigm cases of deontological and consequentialist judgments without already deciding what is essential to them, and then argue that when empirically examined each has in common that they are responses to one of two contrasting factors. Now, in our view Greene is not in fact right about what each has in common (Campbell & Kumar, forthcoming). But the claim can be empirical and not merely armchair. [10] Universalization tests, which Hare (Citation1981) thinks define moral evaluation, are discussed extensively in the literature. "Moral judgments," Hare says, "are universalizable in … that they entail identical judgments about all cases identical in their universal properties" (1981, p. 108). Arguably, consistency reasoning is the application of a more demanding standard, that cases be identical in all their morally relevant properties. [11] In Campbell and Kumar (forthcoming) we argue that the inconsistency at play here is often practical, rather than logical. One emotional response is practically inconsistent with another in the sense that they motivate incompatible behavior. [12] Thomson's argument is very often seen as consistency reasoning, as in the context of the research cited in section 6, and for that reason we use it to illustrate this important form of argument. To be fair, it can, and perhaps should, be interpreted otherwise, as giving a counterexample to a general moral principle, such as "saving an innocent person's life always takes moral priority over personal inconvenience," since saving the life of the violinist in the example appears not to have that moral priority. [13] It's not entirely clear which version of premise (2) Greene accepts in his (Citation2010) work. Sometimes he says things that suggest the version above, sometimes the version we will defend below. How exactly to interpret Greene does not matter much to us. What matters is that in either case his argument fails. [14] A deontological moral theory that endorsed the doctrine of double effect on the basis of its ability to capture intuitions about cases other than trolley cases would not be subject to this debunking argument. [15] In Campbell and Kumar (forthcoming) we endorse a "minimalist" dual process model of moral judgment but argue that this pattern among "deontological" and "consequentialist" judgments is illusory. The former do not derive solely from system 1; the latter do not derive solely from system 2. For similar criticisms see also Kahane et al. (forthcoming). [16] However, what would it be to successfully deploy this criterion? To claim that a moral problem is familiar or new is to stake out a certain normative position, a position that needs to be made explicit. This is an important issue, but we won't pursue it here. [17] Greene has so far paid little philosophical attention to this complication in his published work. He argues, briefly, that the intuitive distinction from which philosophers construct the doctrine of double effect is untrustworthy because, rather than being embodied in an internal norm, it falls out of the structure of our action planning system (Greene, Citation2010, p. 17). This argument runs afoul of the same kind of problems we discussed in section 5. A crucial normative assumption is required: that the way in which a distinction between means and side effects is represented in our cognitive system makes it likely to be unreliable. Again, we are given no reason to think that is true. [18] Sometimes researchers construct minimal pairs in their stimuli, but not in the subject's representation of the stimuli. That is, the stimuli are most easily interpreted in a way that leads to further differences. An important methodological virtue is thus the production of psychologically effective minimal pairs.