EAG4

Journal of Moral Philosophy ^ ^ ^ ^ © 2006 SAGE Publications London, ^ ^ T M ) O U R N A L

Thousand Oaks. CA ^^H^fl OF and New Delhi ^ ^ ^ ^ | MORAL

http://MPJ.sagepub.com J i ^ ^ l PHILO.SOPHY

Vol 3(2): 193-213 DOl: 10.1177/1740468106065492

Virtue, Character and Situation

JONATHAN WEBBER*

Department of Philosophy University of Sheffield Sheffield SIO 2TN, UK [email protected]

Philosophers have recently argued that traditional discussions of virtue and character presuppose an account of behaviour that experimental psychology has shown to be false. Behaviour does not issue from global traits such as prudence, temperance, courage or fairness, they claim, but from local traits such as sailing-in-rough-weather-with-friends-courage and office-party-temperance. The data employed provides evidence for this view only if we understand it in the light of a behaviourist construal of traits in terms of stimulus and response, rather than in the light of the more traditional construal in terms of inner events such as inclinations. More recent experiments have shown this traditional conception to have greater explanatory and predictive power than its behaviourist rival. So we should retain the traditional conception, and hence reject the proposed alteration to our understanding of behaviour. This discussion has further implications for future philosophical investigations of character and virtue.

Keywords: character traits; situadonism; social psychology; virtue ethics

1. Introduction

What is character? What do the distinctive patterns we discern in our own and one another's behaviour consist in? Exactly what does it mean

to call somebody honest, compassionate or courageous, and how are such epithets earned? Answering these questions is central to assessing the various ethical theories that enjoin us to develop morally sound character traits, since we need to know in some detail what character traits really are before we can discern whether and how they can be developed. John Doris has argued that empirical evidence indicates that we do not have characters as these are gen- erally understood in ethical discourse. There are no such traits as prudence, temperance, courage, or fairness, he argues. There are only such traits as sailing-in-rough-weather-with-friends-courage and office-party-temperance,

* For comments on earlier drafts, I am very grateful to Chris Bennett, George Botterill, Paul Faulkner, Chris Hookway, Sarah Parker, Tom Simpson, James Tartaglia, Tom Walker, Suzi Wells, an audience at Keele University, and two anonymous referees for this journal.

194 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

which are perfectly compatible with acting in a cowardly fashion when sailing in rough weather with new acquaintances and drinking excessively at family gatherings. Each trait is to be specified with reference to a range of features of the situations in which they are manifested, he claims, with the result that each person has a wide range of traits each with very restricted situational application. Following his terminology, we can refer to this as the 'fragmen- tation' theory of character.'

Doris goes on to argue that adopting this fragmentation theory should lead us to 'a certain redirection of our ethical attention': instead of attempting to improve our characters, we should 'invest more of our energies in attending to the features of our environment that influence behavioural outcome'; we 'should try, so far as we are able, to avoid.. .ethically dangerous circumstances' and seek out 'situations conducive to ethically desirable conduct'. The ideal of character-development recommended by many ethicists should be aban- doned in favour of an ethic of situation-management. Maria Merritt has for- mulated a variant of the fragmentation theory, according to which the aspects of situations that need to be built into the concept of each trait are always social aspects. She agrees that we should abandon character-development in favour of situation-management, adding that we should focus on manipulat- ing our social settings and relationships.^

But not all adherents of the fragmentation theory agree. Peter Goldie advo- cates the aim of harmonizing one's fragmentary traits into semblances or simulacra of traits as traditionally conceived. If I lcnow that I am only sociable in certain situations, for example, and I approve of that sociability, then it seems that I would prefer to be sociable generally. With the help of the twin 'executive virtues' of circumspection and strength of will, I can manipulate my surroundings so that they elicit sociable behaviour, and continually doing so will gradually alter my situation-specific dispositions, so that I am disposed to be sociable across a wide variety of situations.^

These recommendations might be criticized for a variety of reasons. One's power to manipulate one's situations will always be limited, so it might seem that the advice of Doris and Merritt does not leave us with enough control over our behaviour. If one's traits are indeed fragmentary, then they will also be vastly more numerous than has traditionally been thought, so Goldie's advice might also seem impractical. But these lines of thought will not be pursued here.''

1. John Doris, Lack of Character: Personality and Moral Behaviour (Cambridge: Cambridge University Press, 2002), chs. 1-5; Peter Goldie, On Personality (London: Routledge, 2004), chs. 3 and 4; the sailing example is taken from p. 115.

2. Doris, Lack of Character, pp. 146-47; Maria Merritt, 'Virtue Ethics and Situationist Personality Psychology', Ethical Theory and Moral Practice 2 (2000), pp. 365-83.

3. Goldie, On Personality, chs. 3 and 4. 4. For development of the idea that the ethical advice issued by Doris and Goldie is

impractical, see my 'Character, Global and Local' (forthcoming in Utilitas).

WEBBER Virtue, Character and Situation 195

The aim of this paper is rather to show that we should not accept the frag- mentation theory of character that these recommendations are based on. The next two sections divide the evidence cited in favour of the fragmentation theory between data that should be set aside and data that needs to be con- sidered. The following section shows that the reliable data provides evidence in favour of the fragmentation theory only if it is interpreted in the light of a behaviourist construal of character traits as dispositions to respond in certain ways to certain kinds of stimuli, and that it is compatible with the traditional theory of character if traits are conceived as dispositions to be inclined with a certain strength to behave in a certain way in response to a certain kind of stimulus. We will see in section five that this latter conception is prominent in Aristotle's Nicomachean Ethics and in recent characterological ethical dis- course influenced by it. Section six presents a pragmatic reason to prefer this conception of traits to the behaviourist one favoured by proponents of the fragmentation theory, and the final section considers some implications of this discussion for future philosophical investigations of character and virtue.

2. Specious Evidence

There are five kinds of experiment cited in favour of the fragmentation theory, all of which are cited by Doris, and two of which are cited by Goldie and Merritt.^ Of these five experiments, as we will see, two should be discounted. There is also a sixth experiment discussed by Doris, but although he has been criticized for taking this experiment as evidence for his theory, he does not actually use it in this way, as we shall see.

One of the experiments to be discounted seems to indicate that people are more likely to exhibit helping behaviour if they have recently had good luck. The subjects of this experiment were unaware that the experiment was taking place. Having made calls from a telephone booth, some found a dime in the coin return slot and others looked but found nothing. Each subject was soon walking behind a woman who dropped a folder full of loose papers and began to gather them up. Of those who had just found a dime, 96 per cent helped her. Of those who had not, just 13 per cent helped.^ All three advocates of the fragmentation theory cite this experiment.'^ The results are taken to challenge

5. Gilbert Harman presents a similar but distinct argument in his paper 'Moral Philosophy Meets Social Psychology: Virtue Ethics and the Fundamental Attribution Error', Proceedings of the Aristotelian Society 99.3 (1999), pp. 315-31. He employs some of the data employed by Doris, and argues that character is illusory and so characterological discourse should be abandoned altogether. 1 cannot do justice to his argument here, but see my 'Char- acter, Common-Sense, and Expertise' (forthcoming).

6. Summary of Alice M. Isen and Paula F. Levin, 'Effect of Feeling Good on Helping: Cookies and Kindness ' , founial of Personality and Social Psychology 2 1 . 3 ( 1 9 7 2 ) pp 3 8 4 - 8 8

Study II. 7. Doris, Lack of Character, pp. 30-32; Merritt, "Virtue Ethics and Situationist Personality

Psychology', p. 366; Goldie, On Personality, pp. 30-32.

196 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

the regularity theory: if behaviour could be influenced so strong]y by such a minor event, the thought runs, then the patterns of one's behaviour would seem to reflect the vicissitudes of one's daily life rather than one's character.

It has been pointed out that repetitions of that experiment have yielded wildly differing results.^ But similar experiments have shown that increased sunshine brings with it larger tips for waiters and an increase in the number people willing to answer questions for a survey, and that having recently been given a cookie increases the likelihood that someone will agree to help another person with a task but not the likelihood that someone will agree to hinder someone else where this would be part of an experiment.' Minor situational variations do seem to have an impact on helping behaviour. But as critics have recently argued, these experiments have involved only relatively trivial helping scenarios. Nobody has shown that such minor situational variations affect the likelihood of responding to someone seeming to be in serious distress.'" We should therefore take these experiments to show only that we should not judge character on the basis of such minor acts of help as gathering up someone else's dropped papers.

The other experiment to be discounted is the Stanford Prison Experiment, cited by Doris. Eighteen males who had volunteered to partake in a study of prison life and were judged to be stable and mature were randomly assigned roles of either 'prisoner' or 'guard'. Early one Sunday morning, each 'prisoner' was arrested by real police officers with no warning, as family and neighbours looked on, and was charged with public order offences, taken to a real police station, warned of his rights, fingerprinted, and then transferred blindfold to what he was told was the Stanford County Jail. He was searched, stripped, sprayed with a delousing agent, dressed in a rough smock, and assigned a number to be used in place of his name. A chain was padlocked to his ankle, to be worn at all times, even while asleep. The 'guards' were not trained, and were free to do just about anything they thought necessary to maintain law and order. At 2.30am on the first night, they woke the prisoners and made them line up and give their identification numbers. Dissension was met with forced push-ups, sometimes with someone pressing down on the prisoner's back. The prisoners responded by barricading themselves inside their cells. The guards used fire extinguishers to fight through the barricades and remove the supposed ringleaders to solitary confinement. From here the situation continued to worsen dramatically until the experimenters called a halt after six days of the scheduled fourteen."

8. Christian Miller, 'Social Psychology and Virtue Ethics', The Journal of Ethics 7.4 (2003), pp. 365-92, Appendix.

9. See, for example, John Sabini, Social Psychology (London and New York; Norton, 2nd edn, 1995), pp. 307-10.

10. John Sabini and Maury Silver, 'Lack of Character? Situationism Critiqued', Ethics 115.3 (2005), pp. 535-62 (539-40).

1 I. Summarized from the excellent website www.prisonexp.org, created and maintained

WEBBER Virtue, Character and Situation 197

Whatever this might tell us about the conduct of experiments, it cannot provide strong evidence in favour of the fragmentation of character. Doris implies that the behaviour of these volunteers did not conform to their own general behaviour patterns, and that most people would behave the same way in the same circumstances. But the experiment has never been repeated and there were no control groups: the behaviour of these few subjects simply cannot warrant such claims.

Although the experiment involved eighteen subjects, moreover, these did not enter the experimental situation individually but together, split into two rival groups. The implication that all the subjects playing the guard role were themselves over-zealous to the point of violence is therefore unwarranted. It is just as likely that this is true of only one or two ofthe guards, and that the others found it too difficult to challenge him or them. Similarly, there may have been one or two prisoners persuasive enough to convince the others to rebel. With no repetition and no control groups, we cannot draw conclusions about the motivations of the behaviour of individuals in this situation.

Finally, even if it is true that most people will behave unusually in such a situation, or even that most people will behave in the same ways in such a situation, the extremely disorienting nature of the opening stages of the ex- periment for the prisoners and the sheer strangeness and threatening insta- bi]ity ofthe situation faced by the guards make it difficult to be confident of any extrapolation to less extreme situations. Any evidence this experiment might provide for the fragmentation of character, therefore, would be defeated by this concern, and so the data should be set aside.

Doris also draws attention to studies of honesty and of extraversion and introversion among schoolchildren. In one study, more than eight thousand subjects between the ages of eight and sixteen were observed in situations where they had the opportunity to cheat in tests, lie about whether they had cheated in those tests, fake a record of their athletic performance, steal a small amount of unattended money, and so on. The study found a high correlation between the proportion of a group of children behaving dishonestly in one situation and the proportion of the same group behaving dishonestly in another situation of the same type. Children who cheated in an exam, it seems, are likely to cheat in the next exam. But it also found very little con- sistency across types of situation. A child who cheats in an exam seems no more likely than any other child to steal unattended money or fake a record of athletic performance. So among children at least, there seem to be no uni- fled traits of honesty or dishonesty, only narrower traits such as being an exam-cheat or an unattended-money-thief.

Rachana ICamtekar raises a methodological concern about the use of this data in the debate over character, asking whether it is legitimate to make infer- ences about adult character traits from observation of children who might be

by one ofthe organizers ofthe experiment, Philip Zimbardo. See also Doris, Laek of Character pp. 51-53.

198 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

'more impressionable, less committed to particular ideals of conduct, or less integrated than adults'. '̂ But this is simply unfair to Doris, who does not use the data in this way. He is aware that 'children with developing personalities are plausibly thought to exhibit less behavioural consistency than fully formed adults', and discusses these experiments 'not so much for their evidential role as for the interpretive perspective they provide'.'^ In this, he is right. This data might illuminate the development or structure of the trait of honesty, but cannot provide evidence against the idea that adult behaviour is regu- lated by such general character traits as honesty, generosity and courage.''' We will therefore set it aside.

3. Significant Evidence

The three remaining kinds of experiment do provide reliable evidence con- cerning adult character traits. Probably the best known, and certainly the most dramatic, of these is the Obedienee Experiment, cited by Doris. Subjects were assigned the role of 'teacher' through a rigged draw and an actor posing as a 'learner' was strapped into a chair with an electrode attached to his wrist. The subject was seated before a 'shock generator' consisting of a line of switches ranging from 15 to 450 volts in 15-volt increments. Some of these were labelled: 'Slight Shock' at 15 volts, 'Moderate Shock' at 75, 'Strong Shock' at 135, and so on up to 'Danger: Severe Shock' at 360, and 'XXX' at 435. The subject was told to ask set questions in sequence. If the learner gave the wrong answer or no answer, the subject was to issue a shock, starting at 15 volts and increasing by 15 volts each time. The shocks were fake, of course.

But the shocks did not seem fake: the subject was given a genuine shock of 45 volts as a sample, and the learner grunted at 75 volts, complained that the shocks were becoming painful at 120, refused to go on with the experiment at 150, cried '1 can't stand the pain!' at 180, screamed at 270, refused to answer any more questions at 300 and 315, screamed again at 330, and gave no response at all to any of the final eight shocks. If the subject queried the procedure or asked to stop, the experimenter replied politely but firmly: 'Please continue', then 'The experiment requires that you continue', then 'It is absolutely essential that you continue', and finally 'You have no other choice.

12. Rachana Kamtekar, 'Situationism and Virtue Ethics on the Content of Our Character', Ethics 1 14.3 (2004), pp. 458-91 (466 n. 30).

13. Doris, Lack of Charaeter, p. 63. 14. Other objections to the use of this data are raised by Gopal Sreenivasan in 'Errors

About Errors; Virtue Theory and Trait Attribution', Mind I 11.441 (2002), pp. 47-68, and by John Sabini and Maury Silver on pp. 540-44 of their 'Lack of Character? Situationism Critiqued', Ethics 1 15 (2005), pp. 535-62. These criticisms are misplaced, as Doris rightly does not use this data as evidence. See also note 23 below, and my 'Character, Consistency, and Classification' (forthcoming).

WEBBER Virtue, Character and Situation 199

you must go on'. The experiment ended either after the subject had reached the maximum shock or the subject refused to continue after the fourth reply.' ̂

This experiment was repeated with thousands of subjects in various coun- tries across three decades, and consistently around 65 per cent of subjects continued to administer the shocks all the way up to the maximum of 450 volts. Almost all went at least as far as 150 volts. These results remained the same in versions in which the learner complained of the effects the shocks had on his heart.'*

This is not evidence of widespread sadism. The subjects often displayed 'striking reactions of emotional strain' and afterwards often reported signifi- cant levels of stress and nervous tension. They acted against their compas- sionate inclinations, it seems, out of obedience to the experimenter: in a variation where the experimenter was called away and somebody posing as another volunteer took his place, the proportion of subjects continuing to the maximum shock level was reduced to 20 per cent.'^

Not just any Idnd of authority figure will do, as Sabini and Silver point out. The western societies in which these experiments were conducted were not particularly authoritarian: students often disobeyed their teachers' instruc- Uons, for example, and crime was not uncommon; members of the same society asked to predict how the subjects would behave or asked how they themselves would behave generally predict mass disobedience, and the actual results surprise and dismay members of that society.'^

But this does not mitigate against the idea that the subjects were willing to obey authority of a particular kind. The experimenter was not simply an authority in the sense of being in charge, but also in the sense that he seemed well-versed in the ways of psychological experiments. This impression was reinforced by the language he used in response to protests: not 'I would like you to continue', but 'the experiment requires that you condnue', and so on. His expertise also conferred a certain moral authority on him: his instruc- tions indicated what an intelligent and well-informed person thinks morally appropriate in such a situation. This point is supported by the version of the experiment in which there were two experimenters who disagreed at 150 volts over whether to conrinue: all the subjects stopped at that point.'^ Although this is usually referred to as an experiment concerned with obedience, there- fore, what it uncovered might be better described as deference. In the version

15. Summarized from Stanley Milgram, Obedience to Authority: An Experimental View (New York: Harper & Row, 1974), ch. 2.

16. See Milgram, Obedience to Authority, ch, 4 experiments I and 2, and ch. 6 experiments 5 and 8.

1 7. See Milgram, Obedience to Authority, ch. 12 and ch. 8 experiment 13. 18. Sabini and Silver, 'Lack of Character?', pp. 546-47, esp n. 33. See Milgram, Obedi-

ence to Authority, ch. 3. 19. Sabini and Silver, 'Lack of Character?', pp. 545-51, esp. 547 n. 33 and 550-51. See

Milgram, Obedience to Authority, ch. 8 experiment 15.

200 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

where the experimenter is replaced by someone who appears to be another volunteer, this person inherits the experimental set-up and so inherits some but not all of this deference.

The other experiments are less complicated. One of these is the Bystander Experiment, cited by Doris to show that behaviour is significantly affected by the presence of passive bystanders. One version of this experiment is con- cerned with the likelihood of a particular individual helping someone appar- ently in distress. Students were solicited to participate in a study of games and puzzles. Each subject was met by a woman posing as a market research representative, and shown into a room either alone or in the company of an experimental confederate posing as another solicited student. From the room, the subjects could see in an adjoining office a ramshackle bookcase precariously stacked with papers and equipment.

Each subject was given questionnaires to complete, as was the confederate if there was one, while the woman went into the office, and closed the door. For a few minutes, the woman could be heard moving things around in the office, until there was a loud crash and a woman's scream. 'Oh my God, my foot...', she cried. 'I... I...can't move...it. Oh, my ankle. I can't...can't... can't...getthis thingoff me'. Andsoon, for about a minute, after which she could be heard muttering, then leaving the office by another door. All of this was a recording playing in the office, but only 6 per cent of subjects later reported suspecting that this was so. Of the subjects left in the tesring room alone, 70 per cent offered to help the woman, whether by entering the office or by calling out. Experimental confederates posing as solicited students merely looked up on hearing the scream, then returned to the questionnaires. Of the subjects left with a confederate, only 7 per cent offered help.^"

The final set of data is from the Samaritan Experiment, cited by all three proponents of the fragmentation theory. Students at Princeton Theological Seminary were asked to complete quesrionnaires in one building, then walk to another building and give a short talk. When leaving the first building, some were told that they were running late and should hurry, some that they would reach the next location on time, and some that they were running ahead of time and would arrive early. On the way to the second building, sub- jects encountered someone slumped in a doorway, apparently in need of help. The differing religious and moral outlooks avowed on the questionnaires made no staristically relevant difference to whether the subjects stopped to help. Neither did it matter whether a subject was among those asked to talk about the Parable of the Good Samaritan. What was relevant was the degree of hurry the subjects were in. Of those who had been told that they needed to hurry, only 10 per cent stopped to help; of those who had been told that

20. Summary of Bibb Latane and John M. Darley, The Unresponsive Bystander: Why Doesn't He Help? (New York: Meredith, 1970), pp. 57-60. See also Doris, Lack of Character. pp. 32-33.

WEBBER Virtue, Character and Situation 201

they would be on time, 45 per cent stopped; and of those who had been told that they would be early, 63 per cent stopped.^'

4. Interpreting the Data

It is unclear exactly why proponents of the fragmentation theory take this data to be evidence in their favour. Part of the problem is that all three give ambiguous definitions of the position they are opposed to, the understand- ing of character purportedly prevalent in ethical discourse. Doris writes that a trait as traditionally construed would be 'reliably manifested in trait-relevant behaviour across a diversity of trait-relevant eliciting conditions that may vary widely in their conduciveness to the manifestation of the trait in question'. Merritt writes that a trait as traditionally construed will 'reliably give rise to the relevant kind of behaviour, across the full range of situations in which the behaviour would be appropriate, including situations that exert contrary pres- sures'. Goldie writes that'we tend to think of character traits as... stable and consistent in their manifestation in thought, feeling and action across a wide range of different situations... honest people can be relied on to act honestly wherever honesty is appropriate'.^^ What is meant by a situation being 'con- ducive' to a certain kind of behaviour, or that kind of behaviour being 'appro- priate' in that situation?

These could be read as moral terms: the proponents of the fragmentation theory could be taken to be arguing that the data indicates that whether or not one will respond correcdy to a moral demand for compassionate behaviour depends not on whether one is disposed to do so, but on further situational features such as instructions from an authority figure, one's degree of hurry, and the passivity of bystanders. Some of their critics certainly take this to be their argument. Michael DePaul and Christian Miller, for example, have both claimed that the data is irrelevant to characterological ethical discourse, since that discourse does not hold virtues such as compassion to be widespread, but to be ideals for which we should ^

21. Summary of J.M. Darley and C. Daniel Batson, 'From Jerusalem to Jericho: A Study of Situational and Dispositional Variables in Helping Behaviour', Journal of Personality and Social Psychology 27.1 (1973), pp. 100-108. See also Doris, Lack of Character, pp. 33-34; Merritt, 'Virtue Ethics and Situationist Personality Psychology', p. 366; and Goldie, On Per- sonality, pp. 62-63.

22. Doris, Lack of Character, p. 22; Merritt, 'Virtue Ethics and Situationist Personality Psychology', pp. 365-66; Goldie, On Personality, p. 50. Doris provides much the same under- standing of traits in his earlier paper, where he first puts forward his argument against the traditional conception of character: 'Persons, Situations, and Virtue Ethics', NoOs 32.4 (1998), pp. 504-30 (506-507).

23. Michael DePaul, 'Gharacter Traits, Virtues, and Vices: Are There None?', in Bernard Elevitch (ed.). Proceedings of the Twentieth World Congress of Philosophy. IX- Philosopity of Mind (Bowling Green, OH: Philosophy Documentation Genter, 2000), pp. 150-53; Miller, 'Social Psychology and Virtue Ethics', §1V. See also Sreenivasan, 'Errors About Errors', p. 57. Similarly, when Sabini and Silver argue that the low correlations reported in the honesty

202 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

Some of the language Doris employs suggests that this is indeed his argu- ment. In his discussion ofthe Samaritan Experiment, he anticipates the objec- tion that the subjects who did not stop to help were indeed compassionate but also had a strong concern for punctuality, and argues that this cannot be the case since 'the demands of punctuality seem rather slight compared with the ethical demand to at least check on the condition of the confederate'. This seems to assume that the point at issue is whether the subjects in the experiment had the virtue of compassion, whether they habitually responded compassionately to the apparent distress of others when doing so was morally appropriate. Similarly, he describes subjects in the Obedience Experiment as engaging in 'destructive behaviour', and summarizes the results of the By- stander Experiment as showing that '[mjild social pressures can result in neglect of apparently serious ethical demands',^'' It seems that this language has led critics to take his central claim to be that the experiments show that people do not possess virtues, that they do not habitually respond in the right way to particular morally relevant situational features.

But this is not the point of his argument. In the article in which he first presented his view, he is careful to distinguish the 'descriptive psychology' of traits from the 'normative theory' of virtue in the opening pages before pre- senting his central argument in terms of traits rather than virtues, and then considering whether one might argue that such an attack on the traditional notion of traits fails to impugn the notion of virtue employed in ethical dis- course.̂ ^ His book is also generally clear that his argument aims to show not that people do not have virtues, but that they do not have any character traits as traditionally conceived. This is also the claim made by Merritt and Goldie.

That character traits and virtues are indeed conceptually distinct will be illustrated in more detail in the next section. Here it is sufficient to note that the idea that one can be habitually too courageous or inappropriately honest is a philosophical commonplace. We should therefore take the terms 'condu- cive' and 'appropriate' in these definitions of character traits as non-moral, descriptive terms: the proponents of the fragmentation theory take the data to indicate that whether or not one will respond to the apparent distress of another person by trying to alleviate it depends not on whether one is dis- posed to do so, but on situational features such as instructions from an authority figure, one's degree of hurry, and the passivity of bystanders.

It is unclear just why they understand the data this way. Goldie does not formulate an argument, Merritt does so, but only briefly. She claims that the

experiments are perfectly compatible with the idea that some people are consistently honest while non-honest people are inconsistent, they are mistakenly taking Doris to he arguing that there is no consistent virtue of honesty rather than that people generally do not act consistently, whether this consistency he virtuous or otherwise. See their 'Lack of Char- acter?', pp. 542-43.

24. Doris, Lack of Character, pp. 34, 39, 33. 25. Doris, 'Persons, Situations, and Virtue Ethics', pp, 505, 506-507, 511-13,

WEBBER Virtue, Character and Situation 203

data shows behavioural dispositions to be related to such 'ethically arbitrary situational factors' as one's degree of hurry or the presence of passive by- standers, and that the traditional understanding of character should lead us to expect that this is not the case.'̂ ^ But without further elaboration, it is difficult to see why the traditional theory of character should lead us to expect that people do not have commitments to punctuality or to peer-approval strong enough to preclude their offering help in certain types of situation.

Doris is more explicit. He hinges his argument on the claim that: 'If a per- son possesses a trait, that person will engage in trait-relevant behaviours in trait-relevant eliciting conditions with markedly above chance probability p'P We can make sense of this notion of 'chance' behaviour only as a com- parison between the agent and the population at large: if a person possesses a certain trait, then the probability of that person behaving in a way that manifests that trait in a given situation is significantly greater than the prob- ability of a randomly chosen member of the population doing so. But this seems to imply that traits are comparative notions, and that it is incoherent to hold that a whole population could possess any given trait. This notion is therefore distinct from the notion traditionally employed by theories that encourage us all to develop certain traits. But Doris does not want to provide such a comparative theory of traits: he agrees that 'in principle, every indivi- dual in a population could possess a [particular] trait'.^^ It seems, therefore, that his reference to probability should not be taken as a constitutive claim, but as an evidential one: an abnormally high probability that a given person will seek to alleviate the apparent distress of another person indicates the presence of the corresponding character trait, but the trait could be possessed without that high probability being abnormal.

This reading fits well with his remark about the Obedience Experiment that 'personality research has failed to find a convincing explanation of [these] results that references individual differences'. '̂ The traditional understanding of traits should lead us to expect to find the subjects of these experiments behaving in different ways according to their differing traits. Instead, he claims, we find near-uniformity in the Obedience Experiment, since even those who rebelled at some point had seemingly high levels of shock at the experimenter's instruction, and we find that the likelihood that one will help a distressed stranger varies with one's degree of hurry and with the presence or absence of a passive bystander rather than varying from individual to individual.

It seems that Doris interprets the data this way: evidence for the regularity theory would be provided if there were a significant diversity of behaviour among the subjects of these experiments, but there is not; therefore the

26. Merritt, 'Virtue Ethics and Situationist Personality Psychology', p. 366. 27. Ooris, Lack of Character, p. 19. 28. Doris, Lack of Character, p. 19. 29. Doris, Lack of Character, p. 39. He defends this claim at length on pp. 46-51.

204 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

experiments that we would expect to provide evidence for the regularity theory fail to provide that evidence. His argument is therefore the following modus tollens: if we possess character traits as traditionally construed, then these experiments should yield a diversity of behaviour; but they do not, so we do not possess character traits as traditionally construed.^"

It might be objected that at least some of the data does indicate a diversity of behaviour. The Bystander Experiment suggests that 63 per cent of the population will offer help to a distressed stranger in another room in the absence of a passive bystander but not in the presence of one; 7 per cent of the population will offer help in both situations, and 30 per cent will offer help in neither. The Samaritan Experiment similarly suggests that 10 per cent of the population will help a distressed stranger even when late for an appoint- ment, 35 per cent will not do so when late for an appointment but will do so when on schedule for an appointment or when having time to spare, 18 per cent will not do so when late or on schedule but will do so when having time to spare, and 37 per cent will not help at all. But this objection does not chal- lenge the claim that the Obedience Experiment shows a remarkable uniformity among subjects.

A stronger response is to query the notion of character traits employed in the argument. This construes a trait in purely behaviourist terms, as the dis- position to respond in a certain kind of way to a certain kind of stimulus. But a trait might rather be construed as a disposition towards certain behavioural inclinations in response to a particular kind of stimulus. Situations can pre- sent an array of features eliciting a variety of inclinations that cannot all be acted upon. Subjects in the Obedience Experiment, for example, are presented with the competing demands of concern for the wellbeing of the learner and obedience or deference to the authority of the experimenter, and so may have inclinations against administering the shocks, but also stronger inclinations towards obedience or deference. The emotional strain manifested by some of the subjects sweating, shaking, stammering, and even crying as they obeyed the experimenter is evidence of their inclinations against administering the shocks. One's overt behaviour is the result of the relative strengths of one's

30. See also his statement of this argument in 'Persons, Situations, and Virtue Ethics': 'we are justified in inferring the existence of an Aristotelian personality structure when a person's behaviour reliably conforms to the patterns expected on postulation of that struc- ture. In the psychological lexicon, we can say that trait attribution requires substantial cross- situational consistency... If I am right about the experimental data, systematic observation typically reveals failures of cross-situational consistency' (p. 507, italics original). When Sreenivasan objects that Doris mistakes evidence of flaws in our everyday trait-attributions for evidence that there are no traits as ordinarily construed {'Errors About Errors', p. 54), he therefore fails to engage with the central argument Doris provides (but see also p. 56). When Kamtekar claims that such empirical arguments against 'virtue ethics' wrongly suppose that traits must be distinctive to the individuals possessing them ('Situationism and Virtue Ethics on the Content of Our Character', pp. 467-68), she similarly fails to engage with this argument.

WEBBER Virtue, Character and Situation 205

competing inclinations, on this picture, and the differences between the levels at which different subjects ended the experiment reflect differences in the relative strengths of their competing inclinations just as do their differing manifestations of stress.

We might term this understanding of character the 'regularity' theory, since it claims behaviour to be regulated by long-term dispositions to have inclinations of certain strengths to behave in certain ways in response to certain kinds of stimuli, and the patterns we discern in the behaviour of indi- viduals over time to reflect these dispositions. In order to count as a char- acter trait, such a disposition must yield an inclination of about the same strength whenever the subject is confronted with the relevant kind of situ- ational feature where this is described in non-normative terms, though as we shall see in the next two sections this inclination need not necessarily be consciously felt in order to play the explanatory role required of it.

If we employ this regularity theory when interpreting the data, we find the positive evidence in favour of differing character traits that Doris urges us to look for: as we have seen, the subjects of the Samaritan Experiment differ in their inclinations towards helping a distressed stranger and towards punctu- ality; the subjects in the Bystander Experiment differ in their inclinations towards helping an apparently distressed stranger and towards winning or maintaining peer-approval; and the subjects in the Obedience Experiment differ in their inclinations towards obedience to authority and their inclina- tions against inflicting pain.^'

As we will see in the next section, moreover, the regularity theory underlies key aspects of Aristotle's ethics and of some prominent recent works influ- enced by it. The argument against characterological ethical discourse mounted by Doris therefore misses its target: so long as that discourse retains the regularity theory, the data Doris cites will present no threat to it. But this is not yet to show that there is any positive reason to prefer the regularity theory to the fragmentation theory, which will be shown in section six.

5. Character in Ethical Discourse

Various key aspects of characterological ethical discourse involve this notion of a trait as a relatively stable disposition to be inclined with a certain strength towards a certain kind of behaviour in response to a certain kind of

31. As we saw in section 3, the authority involved in the Obedience Experiment is that of an institutional expert and moral guide, not simply that of the person in charge. Sabini and Silver question whether the behaviour of the subjects in the Samaritan Experiment should he described in terms of punctuality, preferring instead to invoke potential embar- rassment at arriving late ('Lack of Character? Sittiationism Critiqued', p. 558). But these explanations seem complementary rather than contrary. Potential embarrassment could enforce one's acting on one's commitment to punctuality, and without that commitment it is difficult to see why somebody would be embarrassed by being late rather than proud of having stopped to help.

206 JOURNAL OE MORAL PHILOSOPHY 3.2 (2006)

situational feature. The wellspring of the tradition, Aristotle's Nicomachean Ethics, contains a number of prominent examples, and descendents of these ideas prominent in recent literature retain the conception of trait they involve, as we shall see.

Aristotle makes it abundantly clear that the kinds of traits that he recom- mends as virtuous are certainly not dispositions to behave in a certain kind of way whenever or almost whenever a certain kind of situational feature is present. He considers such traits to be vices of excess: giving money away at every or almost every opportunity is not generosity, but profligacy; never or hardly ever recoiling from danger is not courage, but foolhardiness; and so on. The vice consists not just in behaving in a certain way too often, but in behaving in that way on inappropriate as well as appropriate occasions. Pos- sessing a specific virtue, on the other hand, means that one will act in a cer- tain way 'when one should, towards the things one should, in relation to the people one should, for the reasons one should, and in the way one should'.^^ Someone might commit murder in the face of real danger, but would not thereby display the virtue of courage, since the intrepid act was inappro- priate for other reasons. Someone who tells the truth even in situations where the demands of compassion make it inappropriate does not fully possess the virtue of honesty.

Underlying this theory is the idea that each trait leads one to be inclined with a certain degree of strength towards certain kinds of actions in certain kinds of situations, and that such traits are virtues only if these inclinations are tempered by other inclinations that constrain the range of occasions on which they will result in action. This explains why Aristotle takes the full possession of one virtue to require full possession of all virtues, and to hold that 'virtue in the primary sense' consists in this possession of full virtue. There is, for Aristotle, a single web of interdependent virtues: full possession of any one virtue means habitually being inclined to behave in a certain way with the right degree of strength in the presence of a certain situational feature, where what is right is relative to the strength of one's other habitual inclinations in response to other possible situational features.^^

John McDowell has argued for a similar position on the grounds that 'virtue issues in nothing but right conduct'.^'' Philippa Foot, on the other hand, has disagreed with such a stringent criterion for virtue, arguing that a trait counts as virtuous so long as it usually issues in right conduct. On the occasions when it does not, it is not 'operating as a virtue' because its mani- festation ought to have been precluded by another trait, which the agent

32. Aristotle, Nicomachean Ethics, translated by Christopher Rowe, with introduction and commentary by Sarah Broadie (Oxford: Oxford University Press, 2002). Quotation is from lI06b21-23; see also 1115bl5-20.

33. Nicomachean Ethics, 1144b32-l 145a2. 34. )ohn McDowell, 'Virtue and Reason', Monist 62 (1979), pp. 331-50. Reprinted in his

Mind, Value, and Reality (Cambridge, MA and London: Harvard University Press, 1998), §2.

WEBBER Virtue, Character and Situation 207

unfortunately lacks. But one can, according to Foot, possess a virtue that does not always operate as a virtue.^^ McDowell and Foot nevertheless agree that the traits in question are dispositions to be inclined with a certain degree of strength to a certain kind of action in response to a certain kind of situ- ational feature; they disagree only over the criteria for such traits to be counted as virtuous. Where McDowell holds that the murderer's trait of courage cannot be virtuous because it is not accompanied by the right level of aversion to killing. Foot holds that it is virtuous so long as it is usually manifested in good behaviour.

Aristotle's discussion of self-control and weakness of will also seems to involve this notion of character trait. The difference between genuine virtue and mere self-control, he claims, is that only the latter involves a struggle to overcome contrary inclinations; the difference between vice and mere weak- ness of will is that only the latter involves such a struggle; and the difference between self-control and wealcness of will is that the inclination one believes one ought to act on wins the struggle in the former case, loses it in the latter. The picture is complicated, of course, by the fact that one might have an additional disposition towards being inclined to do whatever one believes one should do. But even when such a trait is in play, it seems that weakness of will results from the inclination acted upon being stronger than the com- bination of the inclination one believes one should act on and the inclination to do whatever one believes one should do. Self-control results from this combination being stronger than the inclination to behave otherwise.

McDowell has recently echoed this view, claiming that the virtuous person acts for a reason that 'is apprehended, not as outweighing or overriding any reasons for acting in other ways...but as silencing them'.^^ The virtuous per- son pursues the right course of action without consciously considering any other. One might disagree and argue that an action is virtuous even when the right reason is apprehended as more important than reasons for contrary actions, whereas merely self-controlled action involves a genuine struggle to act on the right reason. But it would remain that virtue and vice involve a greater difference between the strengths of competing inclinations than do strength and weakness of will.

This difference between acting as a result of conscious consideration and acting without such conscious consideration reflects the difference between the two ways in which Aristotle and his followers think that traits lead to action. When faced with a situation that seems to call for mutually exclusive responses, these responses reflect different if overlapping sets of one's values and attachments. One will therefore need to deliberate, to consider which of the conflicting values one attaches most weight to, which one is most strongly

35. Philippa Foot, 'Virtues and Vices', in her Virtues and Viees and Other Essays in Moral Philosophy (Oxford: Blackvvell, 1978), §111.

36. Aristotle, Nieomachean Ethics, book 7 chapters 1-10. esp. 1146b31-5; McDowell, 'Virtue and Reason', §3.

208 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

inclined to act upon.̂ ^ But sometimes it just seems to one without conscious deliberation that a certain course of action is the right one to pursue. Aris- totle considered character to be reflected in such instances no less than in the outcomes of deliberation, and this idea is emphasized by McDowell along with Iris Murdoch, David Wiggins, and Martha Nussbaum, among others.̂ ^

An inclination to behave in a certain way need not result directly from a disposition to have that kind of inclination in response to a certain kind of feature, therefore, since it could result from the combination of a number of dispositions elicited by different features of the situation. One may be less strongly inclined to tell the truth to someone who might be upset by it than one would be otherwise, but be inclined to do so nonetheless; one's inclina- tion would therefore result from one's disposition towards a strong inclination to tell the truth in conjunction with one's disposition towards a less strong inclination not to upset people. This resulting inclination could result from conscious deliberation, or from the processing underlying one's perception of the situation. In the latter case, one need not even consciously feel the incli- nation not to upset people.

The ideas surveyed in this section, drawn from Aristotle and prominent recent discussions of virtue involve the same idea of a character trait, one that stands in contrast to the conception employed by the proponents of the fragmentation theory of character. We have already seen that the claim that people differ in character is positively confirmed by the data cited against it so long as we interpret that data using this notion of traits commonly employed in characterological ethical discourse, rather than the behaviourist notion employed by the critics of that discourse. But this leaves open the question of which of these conceptions we should prefer. The next section presents an argument in favour of retaining the traditional notion employed by Aristotle and his followers, rather than replacing it with the behaviourist notion employed by Doris, Merritt, and Goldie.

6. Explanation and Prediction

The traditional conception of a character trait as a disposition towards a cer- tain inclination in the presence of a certain situational feature is preferable

37. See Bernard Williams, 'Internal and External Reasons', in his Moral Luck: Philosophi- cal Papers ;973-/980 (Cambridge: Cambridge University Press, 1981), p. 104. Aristotle's lengthy and scattered thoughts on deliberation are discussed in: Richard Sorabji, 'Aristotle on the Role of Intellect in Virtue', pp. 205-207; David Wiggins, 'Deliberation and Practical Reason', pp. 231-37; both in Essays on Aristotle's Ethics, ed. Amelie Oksenberg Rorty (Berkeley, CA: University of California Press, 1980).

38. Nieomachean Ethics, 1143bl 1-14; see also 1 109bl8-23, 1126b2-4and 1142a27-30; Iris Murdoch, The Sovereignty ofCood Over Other Concepts (Cambridge: Cambridge University Press, 1967), p. 20; McDowell, 'Virtue and Reason', §2; Wiggins, 'Deliberation and Practical Reason', pp. 232-33; Martha C. Nussbaum, 'The Discernment of Perception: An Aristotelian Conception of Private and Public Rationality', in her Love's Knowledge: Essays on Philosophy and Literature (Oxford: Oxford University Press, 1990), esp. pp. 74 and 79.

WEBBER Virtue, Character and Situation 209

to its beliaviourist riva] for ihe pragmatic reason that it provides a deeper ]eve] of exp]anation, and so has greater predictive power. The behaviourist notion explains an action by the trait of responding in a certain way to a cer- tain combination of environmental stimuli, such as sailing in rough weather with friends or passing a distressed stranger while in a hurry. The traditional notion, however, can allow us to explain why that combination of stimuli leads to that behaviour from that person. It does so by referring to a number of inclinations and their relative strengths elicited in that person by that situation. We can therefore predict that person's behaviour in a novel situ- ation that combines only features already observed, by considering the out- come ofthe combination ofthe traits known to be elicited by those features. The behaviourist notion, on the other hand, requires us to consider the novel situation as eliciting a trait not previously observed, and therefore unknown.

Advocates of the behaviourist notion might argue that this deeper level of explanation is simply not available, that we can do no more than observe overt action in response to environmental stimuli. They might further claim that belief in this deeper level of explanation is the reason why predictions of behaviour often fail. One way to decide whether this is true would be to undertake longitudinal studies. If consistent observation and classification of the actions of a few individuals over significant stretches of time were to yield reliable predictions of their behaviour in novel situations, then the traditional notion would be shown to be preferable to the behaviourist notion. If it were to yield no such reliability, then the traditional notion would have no such advantage.

Such longitudinal studies would require long-term detailed surveillance of the public and private lives of people unaware of being under surveillance and are therefore unavailable to professional psychologists for both logistical and ethical reasons,^' But these kinds of studies are not necessary for settling this issue. Latitudinal studies, in which different subjects are tested in the same situation, can provide evidence of this deeper level of explanation, and hence also evidence of character traits as traditionally construed.

One study that provides such evidence was originally designed to explain the following difference in homicide rates between the northern and southern regions of the USA: white males in large southern cities are no more likely to commit homicide than are their counterparts in large northern cities, but those who do commit homicide are significantly more likely to do so as a result of an argument; and white males in rural areas and small urban areas in the southern states are no more likely to commit homicide as a result of a

39. Some philosophers argue that longitudinal studies are available in the form of novels and the writings of other acute observers of human behaviour. I defend this view in 'Char- acter, Common-Sense, and Expertise'. But since the traditional notion of a character trait can be shown to be preferable to its behaviourist rival by reference to recent experimental data, there is no need to make a more controversial appeal to non-experimental literature here.

210 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

crime than are their northern counterparts, but are twice as likely to commit homicide as a result of an argument.'"^

The experimenters tested the hypothesis that these differences were due to a southern culture of honour, more pronounced in rural and small urban areas than in larger, more cosmopolitan cities. Their subjects were white, non- Jewish, non-Hispanic, male citizens ofthe USA studying at the University of Michigan. They were divided between 'southerners', defined as having lived in 'southern' states for at least six years, and 'northerners', who had lived in 'southern' states for less than six years if at all. On average, the southerners had lived in 'southern' states for 80 per cent of their lives, the northerners for 5 per cent.''' Evidence of a southern culture of honour would be provided if southerners were shown to he significantly more likely than northerners to react violently to insults.

Each subject was asked to complete a questionnaire, deliver it to a room at the end of a long corridor, and return. Some subjects passed someone in the corridor who had to close a filing cabinet to let them pass, then had to do so again on their return, and so bumped them with his shoulder and insulted them as they passed. For control subjects, the corridor was empty. There are three variations of the experiment. In one, subjects were then asked to com- plete a story, which began at a party with Jill telling her fiance Steve that a mutual friend Larry, who knew that Jill and Steve were engaged, had been making passes at her, and stopped with Steve seeing Larry trying to kiss Jill. 75 per cent of insulted southerners ended the story with Steve injuring or threatening to injure Larry, whereas only 20 per cent of control southerners did so. Having recently been insulted made no statistically relevant differ- ence to how northerners ended the story.

In the second variation, saliva tests were taken before the questionnaires were filled in and after the subject returned from dropping off the question- naire. Differences in levels of two hormones were measured: cortisol, associated with high levels of stress, anxiety and arousal; and testosterone, associated with aggression and dominance behaviour. The cortisol levels of insulted southerners rose by an average of 79 per cent during the experiment, whereas those of control southerners rose by an average of 42 per cent. The cortisol levels of insulted northerners rose by an average of 33 per cent, compared with the control northerners' average of 39 per cent. Similarly, testosterone levels rose 12 per cent for insulted southerners and only 4 per cent for con- trol southerners, where these rose 6 per cent for insulted northerners and 4 per cent for control northerners. So there is a significantly larger average

40. See Richard E. Nisbettand Dov Cohen, Culture of Honor: The Psychology of Violence in the South (Oxford and Boulder, CO: Westview Press, 1996), p. 21.

41. For more on the definitions of 'northern' and 'southern' used here, see Nisbett and Cohen, Culture of Honor, Appendix C.

WEBBER Virtue, Character and Situation 211

increase in both cortisol and testosterone levels during the experiment for insulted southerners than for control southerners or any northerners.

In the third, 'chicken' variation, the subject returning along the corridor was faced with somebody coming the other way, who was six feet and three inches (one metre and 90 centimetres) tall, weighed 250 pounds (17 stone and 12 pounds; 114 kilograms), and played college football. All subjects gave way. They differed in the distance at which they did so. Insulted southerners yielded at an average of around three feet (90 centimetres), control south- erners at about nine feet. Insulted northerners yielded at around six feet, control northerners at around five and a half. The insults therefore made a significant difference in the case of southerners, but not in the case of nor- therners,*^

If we adopt the behaviourist notion of traits, these experiments suggest only that none of the subjects have the trait of responding violently to insults in experimental situations, that southerners are more likely than northerners to have the trait of completing the story violently after being insulted in an experimental situation, and that southerners are more likely than nor- therners to have the trait of displaying increased bravado shortly after being insulted in experimental conditions. These are distinct traits. The data con- cerning cortisol and testosterone levels does not report any behaviour, and so does not indicate any difference in traits. The traits that are indicated, more- over, cannot be linked to any data concerning homicide occurring outside experimental conditions, since the traits manifested are relative to those con- ditions. Indeed, since homicide is committed in a wide range of situations, the behaviourist notion of traits precludes any explanation of the homicide data in terms of trait differences between southerners and northerners.

The traditional notion of traits, on the other hand, allows us to explain the homicide data and the data from the story and chicken variations of the experiment in terms of a trait prevalent among southerners but not nor- therners: a disposition to be strongly inclined to respond violently when insulted. Evidence of the inclination elicited by insults is provided by the saliva tests. The fact that nobody actually did respond violently can be explained by the other relevant inclinations in play in the experimental situ- ation. This information allows predictions of results of future experiments that also test reactions to insults or slights among northerners and southerners, but test these in different ways. The traditional notion of traits therefore genu- inely does have an explanatory and predictive power that the behaviourist notion lacks. We should therefore choose to interpret data using this tradi- tional notion.

42, Nisbett and Cohen, Culture of Honor, ch, A. For further data in support of the theory of a southern culture of honour, see chs, 3 and 5; for more on what this culture involves, see chs, 1 and 6.

212 JOURNAL OF MORAL PHILOSOPHY 3.2 (2006)

7. Conclusions and Implications

At issue between the regularity and fragmentary theories of character is a conceptual question: should we prefer the traditional notion of traits that leads to the regularity theory, or the behaviourist one that leads to the frag- mentary theory? We have seen that the traditional notion is preferable because it allows for a deeper level of explanation, and therefore has greater explanatory and predictive power.

Evidence of this deeper level of explanation is provided by a set of experi- ments that measure not just overt behavioural responses to environmental stimuli, but also inner events mediating between stimulus and response: the saliva tests measure physiological priming for aggression and dominance behaviour, and the story experiment measures cognitive priming for such behaviour; neither measures such behaviour itself. The chicken variation measures subtle dominance behaviour indicative of violent inclinations, but not violent behaviour. The Samaritan and Bystander Experiments, on the other hand, report only stimuli and overt behavioural responses, and the Obedience Experiment measures only these things. There is a methodological lesson in this: if we are to retain the regularity theory of character and the related notion of traits as dispositions to behavioural inclinations, then empirical data employed in philosophical discussions of character should be drawn from experiments reporting inner mediating events, not those report- ing only behavioural responses to stimuli, since inclinations are not always manifest in behaviour. In addition to this, there are further implications for the philosophical discussion of character and virtue to be drawn from the foregoing discussion.

First, the data concerning responses to insults also indicates that the rele- vant dispositions vary with cultural background, which shows that we could in principle inculcate a particular kind of habitual response to insults in our children, at least by manipulating their cultural background. Whether an adult's habitual response to insults can be altered is another issue. Perhaps it is impossible, perhaps it requires relocation to another culture, or perhaps it is possible without relocation. Perhaps it is possible but extremely difficult, either with or without relocation. This is an empirical issue. Whether adults can be enjoined to inculcate in themselves a particular kind of habitual response to insults, and if so how this might most expediently be done, are therefore matters requiring further empirical investigation.

It may be, of course, that some traits do not vary with cultural background, while others do. Perhaps our levels of obedience or deference to certain types of authority, for example, are hard-wired. If so, ethical theories enjoining inculcation of any contrary inclination would be unacceptable, though ones urging us to resist inclinations issuing from such hard-wired traits would not. Ethical theories of character-development would also have to consider the behaviour likely to result from the combination of each recommended trait

WEBBER Virtue, Character and Situation 213

with this set level of obedience or deference to certain types of authority. Fur- ther empirical research is required, therefore, to discern whether any traits are inherent and unchangeable.

Where the recent philosophical literature on ethics and empirical psychol- ogy has discussed character in general, moreover, it seems necessary to treat traits severally. Perhaps concern for peer approval is unalterably high in humans, for example, where concern for the welfare of others is culturally variable. Perhaps concern for punctuality is very difficult to change in adult life, but concern for keeping one's word is not. Ethical injunctions and their implementation will need to take account of such differences between traits. But this is not so straightforward as it might seem, since there seems no obvi- ous reason to believe that the folk psychological terms employed in ordinary English individuate traits correctly. If traits are dispositions towards certain inclinations of certain strengths in response to certain environmental stimuli, their individuation requires reference to the range of stimuli that elicit them. Discussion ofthe individuation of traits should also draw on empirical inves- tigation.

Philosophical discussions of the ethics of character therefore require a considerable amount of empirical information. Two ofthe proponents ofthe fragmentation theory have also argued, as we have seen, that the project of character-development should be replaced with one of situation-management. It might turn out that this advice is right, even though the fragmentation theory of character need not be embraced. It might simply be that character- development is not feasible, either generally or with respect to certain traits. Whether or not situation-management is preferable to character-development, therefore, is an issue to be settled by the same kinds of empirical investi- gation involved in the discussion of the individuation and development of behavioural inclinations.

This is not to deny, of course, that ethics is an irreducibly normative dis- cipline. It is not to suggest that psychology is a guide to moral value. It is just to point out that since a practical ethical recommendation is acceptable only if it is reasonable to expect people to follow it, theories that recommend character-development should be formulated and assessed with reference to empirical studies of the nature of behavioural inclinations, studies that are becoming increasingly available.