In Varieties of Moral Personality, Owen Flanagan articulates a thesis he calls the Principle of Minimal Psychological Realism:
Arpaly (2007, ch. 2) argues by parity with norms for belief, which is commonly not taken to be under volitional control.
261 The classic discussion is Wolf 1982, who argues that moral perfection
“does not constitute a model of personal well-being which it would be particularly rational or good or desirable for a human being to strive” (Wolf 1982, 419).
262 For an overview from a philosophical perspective, see Tiberius 2006.
138
Make sure when constructing a moral theory or projecting a moral ideal that the character, decision processing, and behaviour prescribed are possible, or are perceived to be possible, for creatures like us. (Flanagan 1991, 32)
Theories that fail to meet the constraint articulated by the principle arguably fail at a central task of ethical theory, namely giving us some guidance in decision-making and evaluating. Certainly, many moral philosophers believe that moral theory is not just an attempt to describe and explain how things are in the way that biology or physics is. It also has a practical import. Aristotle saw this clearly:
The purpose of this inquiry is not, as it is in other inquiries, the attainment of theoretical knowledge: we are not conducting this inquiry in order to know what virtue is, but in order to become good, else there would be no advantage in studying it. (NE 1103b) From a more theoretical perspective, minimal psychological realism is supported by the connection between the ability to follow the demands of morality and fairness of holding one responsible for failures, as noted above. There are thus both practical and theoretical reasons for normative ethicists to ensure that they do not set up ideals that would be either cognitively or motivationally too taxing for us.
On the cognitive side, the constraint of psychological realism is widely accepted – just about any act-consequentialist acknowledges that as a decision procedure, it would require “an impossible amount of attention to one’s action options”263. We simply cannot calculate and compare the outcomes of all possible actions. The standard response is to defend act consequentialism as a criterion of rightness instead, and allow for one heuristic or another to do the job of guiding decision.264 Motivational constraints are more controversial. One reason for this is that it is not so clear what we can
263 Flanagan 1991, 34.
264 See for example Hare 1981 on the levels of moral thinking and Railton 1984.
139
and cannot do in this respect. Flanagan distinguishes between natural and social limits on psychological traits.265 Sexual desire, for example, is natural, but its expressions in action are heavily influenced by culture. Consequently, the sort of things that are not too demanding for members of one culture may be such for those who have deeply internalized the practices of another. They may not be strictly speaking impossible, but too costly to be worth it. This is, of course, controversial, since it involves modifying “ought implies can” into something like “ought implies can without excessive cost to personal well-being or self-deception”. Many are tempted to rather conclude with Kant that if we are too weak to meet the demands of morality, so much worse for us: what is wrong is still wrong.
I need not take a position on the motivational demandingness issue here. Instead, I want to discuss empirical evidence for two psychological claims that would relatively uncontroversially require us to modify our normative theories if they were true: psychological egoism and the thesis that there are no such traits as virtues.
The Reality of Altruism
I have already discussed the issue of the motivational effectiveness of moral reasoning and moral sentiments above, so I will limit myself here to the more general question of the possibility – and actuality – of altruism, which provides a good illustration of the different roles of a priori reflection and experimental study. The form of altruism at issue is psychological altruism, which is a matter of having non-instrumental motives to benefit others – that is, motives that do not derive from some concern for one’s own good.266 It must be distinguished from behavioural altruism, which is the rather trivial
265 Flanagan 1991, 41–46.
266 For a definition of psychological altruism that clearly distinguishes it from other things (sometimes misleadingly) called altruism, see Joyce 2005, 13–16.
140
claim that we sometimes act in ways that benefit others at a cost to ourselves267, and from evolutionary altruism, the thesis that sometimes genes that dispose their vehicles (individuals) to sacrifice their reproductive fitness for others get selected for. Opposed to psychological altruism is psychological egoism, the thesis that all our motives are ultimately self-regarding.268 The truth or falsity of altruism is naturally of significance for minimally psychologically realist moral theories, since if it is the case that we are incapable of genuinely other-regarding motives, morality cannot demand them from us, supposing that ought implies can. This would require a radical revision of at least Kantian and virtue theories, as well as some consequentialist theories of agent-evaluation.269
Since psychological egoism and altruism are hypotheses about actual human motivations, they are straightforwardly an empirical matter.270 Testing them empirically is nevertheless challenging, since there is always potentially a gap between external behaviour and ultimate motives, as well as between our introspective view of our motives and our real motives. Getting around this requires considerable ingenuity, and is not easily accomplished. Perhaps the
267 There are forms of behavioural altruism, however, whose existence is contentious. Ernst Fehr and his colleagues in the field of experimental microeconomics have conducted a series of experiments to study altruistic punishment, which is a form of behavioural altruism that seems to play a particularly important role in social coordination. See Fehr and Fischbacher (2003) for an overview of this research, which corrects some simplistic behavioural assumptions that rational choice theorists are sometimes liable to make.
268 As Stich, Doris, and Roedder (MS) remind us, it is not simple to decide which desires are self-regarding and which are not. This is clearly a conceptual question that must be settled before any empirical investigation.
269 Insofar as consequentialists assess actions only in light of their consequences, the possible falsity of altruism does not affect them.
270 Philosophers have, to be sure, not always thought so. Mill 1863, for example, tries to prove a priori that altruism is “physically and metaphysically” impossible because of the connection between pleasure and desire. Blackburn 1998 gently corrects him with Butler’s point about the dependence of many pleasures on pre-existing desires.
141
cleverest experiments to date have been conducted by Daniel Batson and colleagues. Batson notes correctly that altruistic motivation for helping is compatible with receiving psychological and non-psychological rewards (such as pleasure, fame, and money) for it, as long as the rewards are a by-product (a perhaps anticipated but unintended consequence) rather than the ultimate goal of helping behaviour.271 When motivation is egoistic, helping the other is only a means to some benefit for the self. To settle the question about psychological egoism we must thus be able to distinguish between goals, means, and unintended consequences of actions on the basis of observable behaviour.
How, in general, do we do this? Well, goal-seeking behaviour presumably terminates when the goal is reached or believed to be unachievable, assuming the agent is minimally rational. In contrast, if something is a by-product, bringing it about or believing it impossible makes no difference to the behaviour in question.
Suppose Joan’s boss is a big classical music aficionado who goes to all concerts of the local symphony orchestra. One day she announces she is going to go to see a Brahms violin concerto. Joan knows that if the boss sees her there, he will be pleased and think favourably of her in the next round of promotions. How can we discover whether Joan is going to the concert for the music (in which case pleasing the boss is a by-product of her action) or to please the boss (in which case going to the concert is a means to pleasing the boss)? It is hard to be certain, but if finding out that the boss will not be there after all leads Joan to drop her plan, or if finding an easier way to gain promotion has the same result, we can reasonably infer that going to the concert was for her at least in part a means for gaining the boss’s favour, rather than something she wanted to do for its own sake. If she does go anyway, that does not, however, yet show that her ultimate goal in going is listening to music – there could be some further end to which going to the concert is a means. But if no plausible candidate is on offer, we have grounds to conclude that she wants to listen to Brahms for its own sake.
271 See Batson 1991, 64–67; Sober and Wilson 1998, 217–222.
142
In general, if both A and B are known consequences of x’s action, we can reasonably infer that bringing about A is not a means to B for x if x brings A about while believing it cannot lead to B or while believing there is an easier or more efficient way C to gain B.272 This schema is important for testing psychological egoism, since the egoist’s claim is that benefiting others (paradigmatically relieving the suffering of others) is never the ultimate goal, but always the means to some benefit to oneself. Thus we can test various egoist hypotheses by removing the postulated benefits to self or adding an easier way to get them, and then observing whether the subject goes on nonetheless to provide benefit to others. This method is necessarily inconclusive, since there could, in principle, always be some other benefit to self that the action gives rise to. But given general assumptions about human psychology, the number of plausible egoistic hypotheses is very limited. Batson and his colleagues have focused on three most common ones. The alternatives can be diagrammed as follows (simplified and modified from Batson 1991, 76):
272 More precisely, we can infer that bringing about A is not mere means to B. As Aristotle already noted, it is possible to want something for its own sake and for the sake of something else; this is the relationship of virtues to eudaimonia.
Perception of
Each story begins with the perception of someone else in need – they are, after all, alternative psychological hypotheses about the motivation to benefit others. The three first alternatives are egoistic, since in each case, helping another is only a means to some benefit to self. If rewards and punishments are understood as external, the first two hypotheses are very implausible – people certainly seem to benefit others without expecting an external reward for doing so or any kind of punishment for failing to do so. But there are also internal rewards and punishments, like empathic joy, guilt, and shame. These either cannot be had, or are most likely to be had, when we empathize with another person. Empathy, as such, can thus be a source of egoistic motivation.
Batson labels the hypothesis that empathy leads to helping another as a way of getting an internal reward the empathy-specific rewards hypothesis, and the view that empathy leads to helping another as a means to avoid an internal punishment like guilt the
143
144
empathy-specific punishments hypothesis.273 The third hypothesis differs from the second in that what is avoided by helping is simply the unpleasant feeling caused by observing or thinking another’s suffering, not the unpleasant feeling resulting from one’s own failure to respond to it. This is known as aversive-arousal reduction hypothesis. For these egoistic hypotheses, empathy must be understood broadly to cover feelings aroused either by the feelings of others or the objective situation of others when it is such as to give rise to feelings in them. The first disjunct covers sharing the feelings that the other has (like feeling sad that John lost his job because John feels sad that he lost his job), while the second, which Sober and Wilson call ‘sympathy’, involves a feeling for the other that she does not necessarily have herself (like feeling sad for Elisa because her brother has died, even though she does not even know it yet).274 The fourth alternative Batson labels the empathy-altruism hypothesis (EA).
According to it, empathy (which is here narrowly understood as
“feeling sympathetic, compassionate, warm, softhearted, tender, and the like”275) can motivate us to have benefiting another as our ultimate goal. In empirical testing, it functions as the baseline against which egoistic alternatives are tested. I will next briefly summarize some of the experimental results that suggest EA fares better than its egoistic competitors.
According to the aversive-arousal reduction hypothesis (AAR), we are motivated to help another in need, because we feel bad when witnessing someone else in need, and helping is a means to get rid of this bad feeling. The bad feeling is a kind of empathic arousal, since it results from the perception of another’s need. AAR gives rise to a testable prediction. If the real reason for helping is getting rid of empathic aversive arousal, one will take an alternative means to that end (for short, escape) if one is available. More precisely, one can be expected to help at a high rate in an experimental situation if there is no escape, particularly if empathetic arousal is high. However, if there is an easy way to escape, one can be expected to take it whether
273 For Batson’s terms, see Batson 1991, chapter 7.
274 Sober and Wilson 1998, 232–235.
275 Batson 1991, 86.
145
empathetic arousal is high or low. By contrast, if EA is true, the ease of escape should make little difference to helping behaviour, since helping is not a means to reduce one’s own arousal.
To test these predictions, one needs to be able to manipulate both the subjects’ level of empathy and ease of escape. Batson and colleagues came up with the following sort of experimental setup to do this.276 Participants are told that the object of the study is task performance in aversive conditions and its effects on the attitudes of others. Aversion is to be created by giving electric shocks to the
‘worker’. At the beginning of the experiment, participants draw lots for the roles of worker (in fact, a confederate) and observer; the draw is rigged so that the subject always receives the role of the observer, watching the worker, Elaine, perform ten two-minute number-recalling tasks via television (in fact a pre-recorded videotape). After the first few trials, Elaine begins to look very distressed, and an assistant, Martha, goes to ask her what is wrong. Elaine explains that as a child she was thrown off a horse onto an electric fence, and was told that she might react strongly even to mild shocks in the future.
Martha offers her a chance to quit, but she says she knows the experiment is important and wants to finish what she started. Then Martha hits upon an idea: maybe she could ask the observer to trade places with Elaine, and Elaine reluctantly agrees. A moment later, the experimenter comes to the observation room and tells the subject that Elaine is “finding the aversive conditions pretty uncomfortable”
and that Martha was wondering whether the observer would like to trade places with Elaine, emphasizing that she has no obligation to do so, having drawn the observer lot.
The participants are in four different conditions: easy or difficult escape and high or low empathy. In easy escape, they are told before the experiment that they need to observe only two out of the ten trials, and after Elaine’s distress that if they decide not to trade places, all they need to do is answer a few questions about their impression of Elaine and they may go. In difficult escape, they are told in advance that they have to watch all ten trials, and are reminded
276 Batson 1991, 113–117.
146
again of this commitment when they decide whether to trade places.
Thus, they can either go and forget about Elaine (easy escape) or watch her suffer (difficult escape). To create the empathy conditions, Batson and colleagues make use of previous studies indicating that people are likely to empathize with people who they think are similar to them.277 To manipulate this, the subjects, who have several weeks earlier filled in a personality and values questionnaire, are handed before the experiment a copy of the same questionnaire allegedly filled in by Elaine. In high empathy condition, Elaine’s answers are very similar to the subjects’ own answers, in low empathy, very dissimilar.
The results can be summed up in the following table278 indicating how many percent of the subjects in each condition agreed to trade places and take shocks for Elaine:
Low empathy (dissimilar victim)
High empathy (similar victim)
Easy escape (told to
leave) 18% 91%
Difficult escape (must stay and watch)
64% 82%
Recall that it follows from AAR that people would prefer an easy escape to costly helping (here receiving electric shocks), since helping is only a means to reducing one’s own aversive arousal. So the proportion of helpers should be smallest in the easy escape/high empathy condition. But the results are the opposite: people with high empathy are most likely to help when they have an easy
277 In another similar experiment, empathy was manipulated by placebo pills. In studies that were based on the subjects hearing a fake radio broadcast and being offered a chance to help the person in talked about, empathy was manipulated by asking participants in advance either to give an objective description of the elements of the story (low empathy) or think about what it is like to be the person talked about (high empathy).
278 Based on Batson 1991, 116.
147
alternative means of reducing their own discomfort. This fits with EA, which thus receives corroboration from the experiment.
Of course, this does not suffice to reject psychological egoism or even AAR – perhaps the manipulations fail in some way (for example, escape does not reduce aversive arousal or similarity does not increase empathy). Batson and his colleagues have therefore, first of all, varied the AAR experiments in several ways, such as manipulating empathy by placebo pills or perspective-taking instead of similarity, but the results have been the same – if people empathize with others, they choose to help rather take the easy escape.279 One frequent criticism has been that perhaps the escape is not so easy: after all, one might reasonably feel guilty for walking out on someone receiving electric shocks. This criticism is in effect the empathy-specific punishments hypothesis (ESP). It predicts that people are more likely to help others when they would feel guilty otherwise. To test it, one must therefore create conditions that are otherwise similar but differ in the likelihood that guilt is aroused.
Of course, this does not suffice to reject psychological egoism or even AAR – perhaps the manipulations fail in some way (for example, escape does not reduce aversive arousal or similarity does not increase empathy). Batson and his colleagues have therefore, first of all, varied the AAR experiments in several ways, such as manipulating empathy by placebo pills or perspective-taking instead of similarity, but the results have been the same – if people empathize with others, they choose to help rather take the easy escape.279 One frequent criticism has been that perhaps the escape is not so easy: after all, one might reasonably feel guilty for walking out on someone receiving electric shocks. This criticism is in effect the empathy-specific punishments hypothesis (ESP). It predicts that people are more likely to help others when they would feel guilty otherwise. To test it, one must therefore create conditions that are otherwise similar but differ in the likelihood that guilt is aroused.