I. INTRODUCTION
Scientists and empirically minded philosophers have developed empirically informed ethics to the point where it can no longer be ignored. Particularly troubling are the arguments put forward by Joshua Greene, who conducted a series of experiments which, according to Greene, advance normative ethics by revealing the inner workings of moral judgements. However, Greene makes an even stronger claim based on his empirical research, namely that we may have less confidence in some moral judgements once we understand these inner workings. He argues that the reliability of deontological moral intuitions is severely undermined by empirical findings.Footnote 1 Greene's empirical findings and the arguments based on these have attracted much attention.Footnote 2 This article does not challenge the empirical side of his work; rather, it targets the arguments based on this research. Greene's not always very clearly explicated arguments have led to a number of rebuttals of what authors took to be his position. It is therefore my primary aim to make his arguments more explicit. Only then will we be able to assess their merits, which is my secondary aim. My third and final aim is to determine the broader philosophical relevance of Greene's arguments.
A brief outline of Greene's empirical research (section II), which also clarifies some of his terminology, lays the ground for the search for the arguments based on this research. I begin this search with the Argument from Morally Irrelevant Factors (section III) and proceed with Greene's more recent arguments for two distinct claims. The first claim is that empirical research can be normatively significant. I reconstruct the argument for this claim as the Personalness Argument (section IV). This argument is sound but it has a somewhat limited scope. In order to broaden the argumentative scope, Greene recently offered the Indirect Route Argument, which I discuss in section VI. This argument overcomes some of the earlier one's limitations. However, it is poorly formulated and invalid. Greene has also presented the Combined Argument for his second claim, which is a claim against deontology. Yet, as I argue in section VI, it is not an argument that will trouble too many ethicists working within the traditional understanding of deontology. In the concluding section VII, I argue that the broader philosophical relevance of Greene's arguments is not that they undermine or support certain first-order moral theories, but that they shed light on the ways deontologists or consequentialists typically argue for their respective theories (second-order morality).
II. GREENE'S EMPIRICAL WORK
The starting point of Greene's research was the well-known trolley scenario.Footnote 3 Consider the Bystander Case: a trolley is heading towards a group of five people working on the tracks. The workers will inevitably die if the trolley proceeds on its course. The only possible alternative is to hit the switch next to you that will move the trolley onto a side track, where it will cause the death of the one person working there. Another version is the Footbridge Case, where you are standing on a footbridge spanning the tracks, in between the trolley and the five workers. The only option to save the five workers is to push a large stranger off the bridge. His body would stop the train but the man would be killed.
Greene's take on the trolley scenario is rather novel. He studied people's intuitive responses to them in a lab setting, drawing, inter alia, on functional magnetic resonance imaging (fMRI). His working hypothesis consisted of three elements. First, the intuition in the Footbridge Case (not to push the large man) is easier to explain with deontological ethical theories; it is therefore a ‘characteristically deontological’ judgement. People's intuitions in the Bystander Case (to hit the switch) are easier to explain with consequentialist ethical theories (‘characteristically consequentialist’ judgement). Second, brain regions primarily active in the Footbridge Case are the ones usually associated with emotions; the consequentialist cases are primarily related to activity in the parts of the brain usually associated with cognition (the so-called dual-process theory). Third, there is a difference in reaction time; characteristically consequentialist judgements take longer than characteristically deontological ones.Footnote 4
Greene claims that his experiments have confirmed the second and third hypotheses, and uses this as a basis for his arguments against the reliability of deontological intuitions. Others have contested the claim that the findings do, in fact, support the hypotheses;Footnote 5 for instance, Guy Kahane and his collaborators have pointed to shortcomings in Greene's experimental design and emphasized the difficulty to test for ‘utilitarian’ judgement.Footnote 6 Nevertheless, Greene has made some refinements to his initial test designs and sticks to his main findings. He explains these findings with an important distinction: characteristically deontological judgements are made in situations which are ‘up close and personal’.Footnote 7 This is, in turn, explained evolutionarily: our ancestors were in danger mostly in situations where a threat occurred within very close range. Moreover, the threat was direct and personal. The enemy had to attack physically from a very close distance. For humans, impersonal or indirect dangers might always have existed (climate change, earthquakes, volcano eruptions, agreements among alien tribes to attack, for example), but, at least until very recently, humans lacked the epistemic powers to detect such dangers, track their causes and adapt to their existence. Characteristically deontological intuitions have emerged evolutionarily as fast and automatic emotional reactions to direct and personal challenges.Footnote 8 Characteristically consequentialist judgements take more time, because the automatic deontological response needs to be overridden.Footnote 9
Let me briefly clarify Greene's understanding of ‘intuition’. Many empirical ethicists base their work on an understanding of moral intuitions as a kind of moral belief about particular cases or types of cases that arise more or less spontaneously or at least without conscious inference. These intuitions might directly translate into judgements or might be overridden by other considerations. According to moral psychologist Jonathan Haidt, a moral intuition is a ‘sudden appearance in consciousness of a moral judgment, including an affective valence (good–bad, like–dislike), without any awareness of having gone through steps of searching, weighing evidence, or inferring a moral conclusion’.Footnote 10
This also seems to be Greene's understanding.Footnote 11 There is no agreed-upon understanding of intuitions.Footnote 12 For instance, Walter Sinnott-Armstrong defines moral intuitions as ‘strong immediate moral beliefs’.Footnote 13 This understanding of moral intuitions differs from Haidt's and Greene's in that it only covers immediate moral beliefs that are ‘strong’ in the sense that ‘believers feel confident and do not give them up easily’.Footnote 14 Let us, for the sake of argument, assume that moral intuitions are immediate moral beliefs, regardless of how strong they are.
A further clarifying note is in order. One of the crucial points in Greene's hypotheses, and of his normative arguments building on these, is his understanding of ‘characteristically’ deontological or consequentialist intuitions and judgements: ‘[W]e'll say that consequentialist judgments are judgments in favor of characteristically consequentialist conclusions (e.g., “Better to save more lives”) and that deontological judgments are judgments in favor of characteristically deontological conclusions (e.g., “It's wrong despite the benefits”).’Footnote 15
Note that this is a functional understanding drawing on, but different from, the traditional philosophical conception.Footnote 16 Deontology is identified as the kind of moral theory that emphasizes moral rules, rights and duties. Consequentialism, by contrast, is understood as the kind of theory that ascribes the moral value of an act depending on its consequences alone.Footnote 17 A judgement is ‘characteristically’ deontological or consequentialist if it is ‘naturally justified’ or easier to explain by one moral theory rather than the other. It is thus possible for a dyed-in-the-wool deontologist to make a characteristically consequentialist judgement and vice versa.Footnote 18 This is not just to say that deontologists make judgements that consequentialists would agree with, and vice versa. Greene's functional understanding has nothing to do with actual consequentialist or deontological reasoning; it is exclusively a claim about the explanatory power of either ethical theory for intuitive judgements. In other words, the claim is that certain intuitive judgements (e.g. hitting the switch in the Bystander Case) are easier to rationalize in consequentialist terms (of maximizing a certain good) and others (e.g. not to push the large man in the Footbridge Case) in deontological terms (of rights and duties).
III. THE ARGUMENT FROM MORALLY IRRELEVANT FACTORS
We are now in a position to examine Greene's arguments against deontology. He holds that ‘[t]here are good reasons to think that our distinctively deontological moral intuitions . . . reflect the influence of morally irrelevant factors and are therefore unlikely to track the moral truth’.Footnote 19 The clearest formulation of the argument is by Selim Berker:
The Argument from Morally Irrelevant Factors
P1. The emotional processing that gives rise to deontological intuitions responds to factors that make a dilemma personal rather than impersonal.
P2. The factors that make a dilemma personal rather than impersonal are morally irrelevant.
C1. So, the emotional processing that gives rise to deontological intuitions responds to factors that are morally irrelevant.
C2. So, deontological intuitions . . . do not have any genuine normative force.Footnote 20
Greene by and large accepts this formulation.Footnote 21 Berker criticized this argument on many grounds. His main point is that the argument only hinges on the correspondence between deontological judgements and personalness on the one hand and between consequentialist judgements and lack of personalness on the other. If this is correct, then the neuroscientific findingsFootnote 22 do not play any role in the Argument from Morally Irrelevant Factors. The emotional or cognitive processing behind all of this is irrelevant for the argument.Footnote 23 Greene's initial answer to this critique was that neuroscience indeed does not play an essential role; yet it is still not normatively irrelevant because it supports the dual-process theory.Footnote 24 As we will see later on, it is quite hard to explicate this ‘essential role’ in Greene's more precise arguments.
IV. THE PERSONALNESS ARGUMENT
Another response to Berker is to emphasize that P1 still is an empirical premise. Greene also took that path and argued that empirical research suggests that personalness is indeed a factor that drives responses to trolley cases. Personalness is here understood as having two prongs: the use of personal force and spatial distance.Footnote 25 Significantly more people find it morally acceptable to kill the large man on the bridge if they do not have to push him, but can simply hit a switch next to the man that opens a trap door so the man will fall onto the tracks. An even larger number of people find this acceptable if the switch is not next to the large man but further away. The general rule is this: the further away one is and the less one has to employ personal force, the more likely one is to find it permissible to give up on one person in order to save five. Here is Greene's Personalness Argument:
Personalness Argument
P1. ‘People's judgments in response to trolley problems are strongly influenced by the presence/absence of personal force.’
P2. ‘The presence/absence of personal force is morally irrelevant to the moral acceptability of actions such as these.’
C. ‘People's judgments in response to trolley problems are strongly influenced by at least one morally irrelevant factor, personal force, and are therefore at least somewhat unreliable.’Footnote 26
One should first note what this argument does not say. Unlike the Argument from Morally Irrelevant Factors, it does not mention certain normative theories like deontology. It does not claim neuroscience's normative significance, but only that of empirical research in general. This would probably include pen-and-paper studies as well as online vignette studies and experiments utilizing virtual reality technologies. In this argument, the claim that personalness is a factor driving moral judgements (P1) is backed empirically; but this backing makes no use of fMRI or other high-technology neuroscientific means. P2 stipulates that personalness is a factor that potentially corrupts judgements in trolley-like cases. Greene takes this normative premise to be almost uncontroversial, which I also tend to believe. If one accepts P2, arguendo, then the Personalness Argument is an example of Greene's claim that empirical research can have normative significance (in the sense that it undermines certain normative standards) and can therefore advance normative ethics. Another point worth noting is that the Personalness Argument is merely an argument within the realm of trolley-like cases. The claim that the same conclusion also holds for characteristically deontological judgements beyond the scope of Trolleyology would require further support.Footnote 27
So, in response to Berker's criticism that neuroscience (in the narrow understanding) does not, in fact, feature in his argument against deontology, Greene offers his Personalness Argument as a defence of the normative significance of empirical research in general. It is a sound argument (when one accepts P2) and shows, by example, that normative ethics can be advanced by empirical research. The Personalness Argument makes the general claim true because, for an existence claim, it suffices to provide one instance of what is claimed to exist. Or, in Greene's characteristic hands-on prose: ‘interesting scientific facts about moral psychology can, when combined with relatively uninteresting normative assumptions, lead us to relatively interesting normative conclusions. That's progress, powered by science.’Footnote 28 But because it is only a single example, what one can learn from this argument alone is somewhat limited.
V. THE COMBINED ARGUMENT
Greene then utilized the Personalness Argument for the claim that empirical research can ‘cast doubt on the reliability of deontological judgments without doing the same for consequentialist judgments’.Footnote 29 He has offered different arguments for this claim. One is the Argument from Morally Irrelevant Factors outlined earlier. A new argument combines the Personalness Argument and the dual-process theory.Footnote 30 The dual-process theory of morality has two important elements: the first is the claim that moral judgements are primarily influenced either by more or less automatic and emotional responses or by more or less controlled and conscious responses. The second element is the so-called ‘Central Tension Principle’, which has it that ‘Characteristically deontological judgments are preferentially supported by automatic emotional responses, while characteristically consequentialist judgments are preferentially supported by conscious reasoning and allied processes of cognitive control.’Footnote 31 Given the qualification of the functional understanding of deontology and consequentialism, Greene holds these two elements of the dual-process theory to be well-supported by empirical research, including neuroscience.Footnote 32 Important for the argument against deontology is the more specific second element, the Central Tension Principle. Combining this principle, the Personalness Argument and Greene's functional understanding of characteristically deontological (or consequentialist) judgements, the following argument can be made:Footnote 33
Combined Argument
P1. In situations like the Bystander Case, people tend to judge in characteristically consequentialist ways, i.e. their judgements are easier to explain by consequentialist theory (and preferentially supported by conscious reasoning and allied processes of cognitive control).
P2. In situations like the Footbridge Case, people tend to judge in characteristically deontological ways, i.e. their judgements are easier to explain by deontological theory (and preferentially supported by automatic emotional responses).
P3. Situations like the Footbridge Case have a high level of personalness.
P4. Situations like the Bystander Case have a low level of personalness.
P5. People's judgements in response to trolley problems are strongly influenced by the level of personalness.
P6. The level of personalness is morally irrelevant.
C1. In situations like the Footbridge Case, people's judgements are strongly influenced by the level of personalness.
C2. In trolley-like cases, characteristically deontological judgements are strongly influenced by at least one morally irrelevant factor.
One point worth noting is that, although the Combined Argument is primarily an argument against the reliability of characteristically deontological judgements, it is also supposed to show that a broader range of empirical findings than those that figure in the Personalness Argument can yield normative implications (thereby also being an example for the normative significance of empirical research).Footnote 34 More specifically, the Combined Argument aims to show that cognitive processing (system 1 vs system 2) influences moral judgement in morally relevant ways.
Another point is that the Combined Argument, as far as its normative conclusions are concerned, is merely an argument within the realm of trolley-like cases, because it builds on the Personalness Argument. Note also that the Combined Argument is merely an argument against characteristically deontological judgements, not against deontology in the traditional philosophical understanding. This functional understanding of deontology and utilitarianism can lead to misunderstandings and makes the translation of the empirical findings into philosophical language rather difficult.Footnote 35 There are two main obstacles for translating Greene's findings back into philosophical parlance. One, by definition, his functional distinction does not cover all deontological or consequentialist judgements, but only those that are easier to explain by one of them. All those judgements which are as easy to explain by one theory as by the other are not covered. Two, Greene's functional distinction might well count judgements as ‘characteristically’ consequentialist (or deontological) that no contemporary consequentialist (or deontologist) would arrive at when applying her theory, because Greene's distinction rests on nutshell versions of both theories, which hardly resemble the elaborated theories actually defended by contemporary ethicists. To give but one example, here is a contemporary version of a rule-consequentialist moral principle: ‘An act is wrong if and only if it is forbidden by the code of rules whose internalization by the overwhelming majority of everyone everywhere in each new generation has maximum expected value in terms of well-being (with some priority for the worst off).’Footnote 36
Accompanied by further qualifications and meta-rules in the overall moral theory, this version of consequentialism is so sophisticated that it hardly fits into Greene's simplifying dichotomy (maximizing a certain good vs rights and duties talk). Greene does not offer an explanation of how to translate his conclusion back into traditional philosophical parlance.
Critique
Greene's argument against deontology is problematic for several reasons. First, recall that Greene offered the Combined Argument as a response to Berker's critique that neuroscience does not play any role in the Argument from Morally Irrelevant Factors. However, the Combined Argument does not feature the neuroscientific elements of the dual-process theory either. The argument does mention automatic emotional responses and cognitive control but only in parentheses in P1 and P2. As with the Argument from Morally Irrelevant Factors, the conclusions of the Combined Argument only hinge on the correspondence between characteristically deontological judgements and personalness; neuroscientific findings do not play any role in the argument. The argument works without the parentheses in P1 and P2. The Combined Argument is thus not a good response to Berker's critique. It is thus also not another example of the normative significance of empirical research. However, this critique does not undermine the argument's force. Its conclusions might still be warranted, although neuroscientific findings are not necessary for the argument's success. One should note though that the remaining core of the Combined Argument is merely a specification of the Personalness Argument plus the empirical premises P1 and P2 without the content added in parentheses.
Second, the Combined Argument does not seem to match with some of the cases and illustrations Greene has discussed or even used in his experiments. Among these is Peter Singer's well-known discussion of the moral obligation of citizens in affluent nations to relieve famine in faraway places in the developing world.Footnote 37 How is it possible that many find it unacceptable to allow a child to drown in a nearby pond even though helping the child would muddy up their new suit, when they also believe that it is acceptable to spend one's disposable income on luxury products even if this money could effectively feed a starving child in the developing world? In both cases, help is needed and can be offered with only a minor sacrifice. Greene has repeatedly taken up this scenario and argued that deontological intuitions do not do their job in the faraway cases, because they lack personalness; consequentialism, by contrast, allows us to explain why there is no relevant difference between our obligation to help the drowning child nearby and the starving child faraway.Footnote 38 Greene wants to make an argument by analogy with the argument from trolley cases. However, the analogy does not hold. Recall that, according to the Combined Argument, cases with low levels of personalness (such as the Bystander Case) yield characteristically consequentialist judgements. This seems not to hold in the drowning child case. People do not judge characteristically consequentialist when the child is at a distance.
The reason for this discrepancy might be that Singer's case goes beyond the Combined Argument, in that it is not a trolley-like case: it is not a case of counting numbers. One does not have to decide against one person in order to save others. I take trolley cases to be a certain kind of moral dilemma. They are similar to ‘moral emergencies’:Footnote 39 there is a single agent, faced with an exhaustive set of clearly defined options to act, all of which have very serious consequences.Footnote 40 On top of that, trolley cases have to do with counting numbers – a trade-off between high values (such as five lives vs one). So, the Combined Argument does not seem to apply to (and is not analogous to) some of the illustrations Greene invokes, such as Singer's drowning child case, because the latter is beyond the scope of trolley cases to which the Combined Argument is limited.
Third, the dichotomy between characteristically deontological and consequentialist judgements does not, by definition, hold in the drowning child scenario of Singer's case, because in this scenario, both deontology and consequentialism establish the duty to help. This duty is thus not easier to explain in terms of rights and duties than by reference to consequences and, hence, not easier to explain by either deontology or consequentialism (which would be required for Greene's functional understanding). Singer's case is thus so different from trolley-like cases that, contrary to Greene's allegation, the Combined Argument does not match with cases such as Singer's.
Fourth, Greene's claim against deontology faces a more general problem. It expresses a simplistic idea because it mirrors the simple version of act consequentialism he endorses. This consequentialism only deems the consequences of an act relevant, whereas deontology (in Greene's vague characterization) potentially also includes other factors such as intentions, blame, causation or personalness. This form of consequentialism might have the advantage of simplicity in that it has a clear focus and confidently rules out ‘irrelevant factors’. The downside is, of course, that it might be too simple in also ruling out factors that are widely – and probably for good reasons – deemed relevant.Footnote 41 It is therefore not surprising to learn that a theory that takes more features of a situation into account is also more sensitive to a variety of features, some of which may, on reflection, turn out to be irrelevant – just as it is easier to hit an elephant than a mouse.
To sum up: given its severe limitations – that it is merely an argument against deontology in the functional understanding and is restricted to trolley-like cases – the Combined Argument seems to be sound, at least when one accepts the moral irrelevance of a situation's level of personalness. The argument is problematic for various reasons. But all of these reasons concern Greene's analogies, illustrations and suggestions to move beyond the Combined Argument and leave the soundness of the argument unchallenged.
VI. THE INDIRECT ROUTE ARGUMENT
Recall that Greene's Personalness Argument is a sound argument for the claim that normative ethics can be advanced by empirical research. However, its reach turned out to be rather limited. It is for this reason that Greene recently attempted to provide a more general argument for the normative significance of empirical research.Footnote 42 The normative significance meant here is not merely the factual knowledge required to apply given normative standards (such as medical knowledge relevant for the application of principles of medical ethics, say), which is sometimes called ‘empirical ethics’.Footnote 43 Rather, what is meant is that empirically acquired facts or empirically well-supported assumptions can support or undermine certain normative standards without committing the is–ought fallacy. It is in this sense that some speak of ‘empirically informed ethics’Footnote 44 when empirical research is invoked to argue for or against a normative standard.
Due to premise P1 and Greene's experimental design, the Personalness Argument's conclusion only holds for trolley cases, a family with very few interesting members. It might be plausible to assume that further empirical research will reveal other factors beyond personalness that are normatively significant, probably even in situations other than trolley cases.Footnote 45 A problem for this line of research is to find arguments with normative premises that are as uncontroversial as the moral irrelevance of the respective level of personalness. However, as it stands, the Personalness Argument does not generalize beyond the scope of trolley cases and considerations of personalness.
Let us now look at Greene's recent attempt to provide a more general argument for the normative significance of empirical research that aims at overcoming the limitations of the Personalness Argument. Here is the argument:
Indirect Route Argument
P1. Moral judgements are primarily influenced either by more or less automatic and emotional responses or by more or less controlled and conscious responses.Footnote 46
P2. ‘Automatic settings can function well only when they have been shaped by trial-and-error experience.’Footnote 47
P3. ‘Unfamiliar* problems [are] ones with which we have inadequate evolutionary, cultural, or personal experience.’Footnote 48
P4. ‘Knowledge of which judgments are preferentially supported by automatic settings vs. manual mode is precisely the kind of knowledge supported by the research [supporting the dual-process theory (inter alia, cognitive science)].’Footnote 49
C1. ‘When we are dealing with unfamiliar* moral problems, we ought to rely less on [automatic emotional responses] and more on [conscious, controlled reasoning].’Footnote 50
C2. ‘[C]ognitive science gives us normatively significant information’Footnote 51 [From P2, P4, C1].
The Indirect Route Argument is labelled ‘indirect’ because ‘cognitive science gives us normatively significant information – information that can nudge us, if not propel us, toward new and interesting normative conclusions’.Footnote 52 The argument's basis is the dual-process theory (P1). P2 is an elaboration of the evolutionary explanation of Greene's findings discussed above. The idea is that automatic settings only develop well through the genetic or cultural transmission of experience or through personal experience. Just as we have to learn how to drive a car, so too do we have to learn how to act morally if this is to develop into automatic responses. With the ‘automatic setting’ in P2 and P4, Greene alludes to his preferred illustration for his research – a digital photo camera with two complementary settings, automatic and manual.Footnote 53 The automatic setting is optimized for typical situations (landscape, portrait, etc.). It is a quick, simple point-and-shoot setting, whereas the manual setting takes more time because one has to adjust the exposure time, focus, etc. The manual setting is more suitable for unusual circumstances that require more flexibility. Greene believes that the
human brain has the same general design. First, we humans have a variety of automatic settings – reflexes and intuitions that guide our behavior, many of which are emotional. We may be conscious of such emotional responses, but we are generally not conscious of the processes that trigger them. We rely on our automatic settings most of the time, and they generally serve us well. Our brains also have a manual mode. It is a general-purpose reasoning system, specialized for enabling behaviors that serve long(er)-term goals, that is, goals that are not automatically activated by current environmental stimuli or endogenous somatic states. The operations of this system are typically conscious, experienced as voluntary, and often experienced as effortful.Footnote 54
Although ‘setting’ sounds a bit odd when referring to humans, I stick to Greene's terminology in the Indirect Route Argument in order to avoid confusion. The idea is that automatic settings yield ‘more or less automatic and emotional responses’ (P1), and that ‘more or less controlled and conscious responses’ (P1) are not generated by such an automatic setting but by ‘manual mode’ (P4).
P3 introduces yet another technical term. The asterisk marks the technical understanding of ‘unfamiliar’. Greene regards problems for which we do not have well-functioning automatic settings as unfamiliar*. Indicators for unfamiliarity* are that problems stem from recent cultural developments or that we are in deep disagreement over them. P4 is the premise that brings in the normatively relevant empirical research. From these premises (P1–P4), Greene concludes that ‘[w]hen we are dealing with unfamiliar* moral problems, we ought to rely less on [automatic emotional responses] and more on [conscious, controlled reasoning]’ (C1) and, finally, ‘[i]f we believe that we ought to rely on automatic settings vs. manual mode to different extents in different situations, and if cognitive science can tell us when we are relying on automatic settings vs. manual mode, then cognitive science gives us normatively significant information’ (C2).Footnote 55
Critique
This argument is problematic. First of all, C1 clearly does not follow directly from the premises P1–P4. The fact that automatic settings ‘function well’ does not easily translate into an ‘ought’. This leap looks very much like the is–ought problem Greene has been so careful to avoid.Footnote 56 Even for those who do not hold the is–ought divide to be as deep as suggested by the non-naturalists, at least it remains difficult to give a naturalistic account of normative claims. But Greene might just have omitted a bridging premise like this: ‘We ought to rely less on automatic settings in situations where they do not function well.’ This introduction of the ‘ought’ might look agreeable at first. However, what is the reason for the moral ‘ought’ behind the well-functioning? In a related argument, he explicitly says that he takes ‘a biological point of view’ on such value judgements.Footnote 57 This point of view might make sense for the technical understanding of unfamiliarity* that helps formulate an adequate explanation of the responses to test cases. But it does not, in itself, justify a moral ought.
One option would be the pragmatistic assumption that morality has certain functions such as guiding behaviour, fostering confidence-building, cooperation and predictability, and allowing for peaceful conflict resolution. The assumption that such functions are central to all moralities underlies Michael Walzer's idea of ‘minimal and universal moral codes’Footnote 58 as well as universalist moral theories. Greene could argue, then, that ‘functioning well’ means something like fulfilling such universal moral functions and is, therefore, good in a moral sense. However one bridges the argumentative gap between ‘function well’ (P2) and ‘ought’ (C1), this bridging premise is very likely to be very controversial among philosophers. As indicated earlier, part of the appeal of the Personalness Argument is that its normative premise is almost uncontroversial. The much more general Indirect Route Argument lacks that appeal. Its soundness crucially depends on the availability of a bridging premise that is less controversial than the ones just discussed.
The upshot is that this new argument is interesting, because it aims to make a much more general case for the normative significance of empirical research (avoiding the limitations of the Personalness Argument); it also has some initial plausibility. However, the argument takes too indirect a route: it relies on an understanding of ‘functioning well’ that is not explicated. There is a huge leap between this notion and the moral imperative to rely on conscious and controlled reasoning. Moreover, the technical understanding of familiarity*/unfamiliarity* is ambiguous; Greene simply seems to have stipulated that all hard moral problems are unfamiliar*. We can conclude that, in its current form, the argument is invalid. So, unlike the Personalness Argument, even given a charitable reading, the Indirect Route Argument is not very likely to be sound. Greene's ultimate hopes for ‘a more general theory that tells us when our judgments are likely to go astray’Footnote 59 are thus not likely to be fulfilled.
I shall note, though, that another strand of research that draws on similar background assumptions as Greene does for his Indirect Route has gained some traction recently. Some authors take research on moral learning to vindicate some contested moral beliefs;Footnote 60 others, including Greene, remain more sceptical.Footnote 61
VII. PHILOSOPHICAL RELEVANCE
Throughout the article, we have seen that it is very hard to pin down how exactly Greene argues for his claims. Starting from Berker's reconstruction of Greene's Argument from Morally Irrelevant Factors, I have identified four arguments for two distinct claims, namely for the unreliability of characteristically deontological judgements and for the normative significance of empirical research. Concerning the former claim, I argued that the Combined Argument seems to be sound (at least when one accepts the moral irrelevance of a situation's level of personalness), whereas the Argument from Morally Irrelevant Factors is not. Concerning the latter claim, I argued that the Personalness Argument is sound. The Indirect Route Argument, however, is poorly formulated and invalid or, when interpreted very charitably (to avoid the is–ought problem), it is very likely to be unsound. This new argument, thus, does not advance Greene's claim for the normative significance of empirical research beyond the scope of the Personalness Argument. I hope that my reconstructions of Greene's arguments will help avoid some future misunderstandings and thereby enhance the debate. After this philosophical accountant work, let me now, in the remainder of this article, discuss the broader philosophical relevance of Greene's arguments.
The Combined Argument is an argument against characteristically deontological judgements in trolley-like cases only. It will not trouble too many card-carrying deontologists, because the realm of trolley-like cases is rather limited and because there is no apparent way to translate findings about deontological judgements in Greene's functional understanding to deontology in the traditional understanding. Greene admits that ‘cognitive systems . . . do not correspond precisely to distinct ethical schools as designated by contemporary ethicists. . . . Nevertheless, if I'm right, the psychological essence of deontology lies with the automatic settings and the psychological essence of consequentialism lies with manual mode.’Footnote 62 One way to make sense of this claim and of Greene's research more generally is to understand it as focusing not so much on normative theories – such as particular versions of deontology or consequentialism (first-order morality) – but on second-order morality, i.e. the theorizing for or against such first-order theories. So the philosophical relevance of Greene's arguments is not that they undermine Kamm's or Korsgaard's moral theories or that they support Singer's or Hooker's (first-order). It is rather, I suggest, that they shed light on the ways deontologists or consequentialists typically argue for their respective theories (second-order). Let me illustrate this difference with two brief hints at where this second-order theorizing might be applied.
First, consider Micah Lott's recent critique of Greene's arguments against deontology. Lott basically says that when we ask whether an act is morally good or bad, we are not merely asking for an utterance of the first inclination – i.e. intuition – people might have towards this act (which he takes to be Greene's position), but for a normative justification of these inclinations. So
either we can identify some value, or good, or normative standard that will allow us to answer the question about justification in the affirmative or we cannot. If we cannot, then we have not found good [reasons for our inclination and should not rely on it, and vice versa.] And that is true regardless of anything science might tell us about the evolutionary origins of our inclination to condemn.Footnote 63
Lott makes a similar point against Greene's argument regarding the use of moral principles such as the Doctrine of Double Effect (DDE), which is often used to justify responses to trolley-like scenarios. Greene holds that the DDE can be seen as a result of mere intuition-chasing, i.e. a form of ex-post rationalization of intuitions which can be evolutionarily debunked: the DDE ‘doesn't justify the judgments. Instead, the judgments justify the Doctrine. This evidence suggests that the justification for the DDE ultimately comes from nothing beyond the automatic settings that produce the pattern of judgment that it summarizes.’Footnote 64 Lott objects:
We need some consideration – some good/value/norm – that reveals why we are correct (or not) to reason as DDE directs. . . . At this point, however, Greene supposes that the only answer we can give to the justificatory question is pointing to the fact that DDE is a good summary of our intuitions about different cases. And Greene believes – rightly, in my view – that if this is all we can say, then we have an unsatisfying defense of DDE. But what Greene overlooks is that this has nothing to do with automatic or manual mode settings. Rather the important point is that we can ask about the justification for DDE even after we have granted that DDE captures our intuitions about particular cases. For the question remains whether we have good grounds to take these intuitions seriously, to regard them as reliable guides.Footnote 65
Lott's objection is that first-order moral judgements or first-order moral principles such as the DDE are not called into question by empirical findings, but because we have no adequate normative justification for them. This misses two important points. First, as argued above, Greene's successful arguments do indeed call into question particular first-order moral judgements on the basis of empirical findings that are normatively significant. These judgements are not necessarily false, because further reasons supporting these judgements can be offered. But, and this is the second point Lott misses,Footnote 66 when ethicists come up with further reasons supporting their first-order moral theories, the game starts all over again. The new reasons can be subjected to the same kind of empirical inquiry that led to the Combined Argument. Lott is thus right to claim that Greene's arguments do not derail deontology in the traditional understanding, because deontologists can always invoke further justifying reasons that have not yet been debunked successfully. But he is not right to assume that this means that Greene has not found ‘a route from cognitive science to normative ethics’.Footnote 67 As we have seen above, with the Personalness Argument, Greene has shown that empirical research can be normatively significant. Having determined the precise scope of the argument, we can now say that it does not tackle a significant number of potential arguments for first-order deontological theories. It does, however, show that all the further arguments for first-order deontological theories are potentially subject to similar empirically grounded criticism too. This is to say that none of the arguments for traditional deontology (or consequentialism) are immune to this kind of empirically grounded debunking. Ethicists should be open to criticism of this sort and refine or correct their first-order theories when the criticism is sound, and rebut the criticism when it is not. This would simply expand the circle of standard academic exchange of ideas and criticism to include such debunking arguments.Footnote 68 The result would be empirically informed first-order moral theories.
The second brief illustration of second-order theorizing concerns justification methodologies such as Rawls's Reflective Equilibrium, which are standardly used to justify first-order theories of almost all stripes. Take, for instance, Brad Hooker, whose rule-consequentialist principle I quoted above. He explicitly relies on Reflective Equilibrium and says, ‘we search for a coherent set of moral beliefs and are willing to make many revisions so as to reach coherence. But we should start with moral beliefs that are attractive in their own right, that is, independently of how they mesh with our other moral beliefs.’Footnote 69 In reaching for a state of Reflective Equilibrium, Hooker explicitly relies on a number of intuitions he regards as more or less self-evident, offering no deeper theoretical arguments. For example:
Any moral theory will be terribly counterintuitive if it requires you to make every decision on the basis of an equal concern for everyone. To be plausible, a moral theory must leave room for some considerable degree of bias (a) towards yourself and (b) towards your family, friends, benefactors, etc.Footnote 70
Empirically informed arguments in the form discussed earlier could potentially debunk such deeply-held intuitions and thus have normative implications for Hooker's theory, as well as for other theories that employ such intuitions.Footnote 71
These are but two examples for the relevance of Greene's arguments for second-order moral theorizing. Greene's work helps us to better understand the ways deontologists or consequentialists typically argue for their respective theories, and to inquire into the implications these ways might have for the status of their respective first-order moral beliefs.
