PSY22-6a

Research Article

The Role of Conscious Reasoning and Intuition in Moral Judgment Testing Three Principles of Harm Fiery Cushman,1 Liane Young,1 and Marc Hauser1,2,3

1 Department of Psychology,

2 Department of Organismic and Evolutionary Biology, and

3 Department of Biological

Anthropology, Harvard University

ABSTRACT—Is moral judgment accomplished by intuition

or conscious reasoning? An answer demands a detailed

account of the moral principles in question. We investi-

gated three principles that guide moral judgments:

(a) Harm caused by action is worse than harm caused by

omission, (b) harm intended as the means to a goal is worse

than harm foreseen as the side effect of a goal, and

(c) harm involving physical contact with the victim is

worse than harm involving no physical contact. Asking

whether these principles are invoked to explain moral

judgments, we found that subjects generally appealed to

the first and third principles in their justifications, but not

to the second. This finding has significance for methods

and theories of moral psychology: The moral principles

used in judgment must be directly compared with those

articulated in justification, and doing so shows that some

moral principles are available to conscious reasoning

whereas others are not.

A topic of recent concern in moral psychology is the extent to

which conscious reasoning, as opposed to intuition, plays a role

in determining moral judgment (Greene & Haidt, 2002; Haidt,

2001; Pizarro & Bloom, 2003). In terms common to social psy-

chology, the question is whether moral judgment is a controlled

or an automatic process (Bargh, 1999).

The conscious-reasoning perspective has been the central

focus for students of moral development in the tradition of

Kohlberg (1969). Kohlberg characterized children’s moral de-

velopment by focusing on the content of their justifications

rather than the source of their moral judgments. An implicit

assumption of this perspective is that people generate moral

judgments by consciously reasoning over the principles they

articulate in moral justification.

One challenge to the conscious-reasoning perspective comes

from work by Haidt in which subjects failed to articulate suffi-

cient justifications for their moral judgments (Haidt & Hersh,

2001). Haidt (2001) proposed that moral judgments arise as

intuitions generated by automatic cognitive processes, and that

the primary role of conscious reasoning is not to generate moral

judgments, but to provide a post hoc basis for moral justification.

Although there is increasing support for the role of intuition in

moral judgment, some researchers argue that both conscious

reasoning and intuition play a role in judgment, as well as jus-

tification (Greene, in press; Pizarro & Bloom, 2003; Pizarro,

Uhlmann, & Bloom, 2003).

A critical ingredient missing from the current debate is an

experimental method that clearly links data on moral judg-

ment with data on moral justification. Without establishing that

an individual uses a specific moral principle, it makes little

sense to ask whether the content of that principle is directly

available to conscious reasoning. Therefore, in the present

study, we first identified three moral principles used by subjects

in the judgment of moral dilemmas, and then explored the

extent to which subjects generated justifications based on these

principles. Our approach, adopted in part from moral philo-

sophy, was to compare judgments across tightly controlled

pairs of scenarios. We parametrically varied each pair of sce-

narios to target only one factor at a time, holding all others

constant. We use the term principle to denote a single factor that

when varied in the context of a moral dilemma consistently

produces divergent moral judgments. By using the term ‘‘prin-

ciple’’ to refer to such factors, however, we make no assumptions

about the nature of the psychological mechanisms that underlie

sensitivity to them.

Address correspondence to Fiery Cushman, 984 William James Hall, 33 Kirkland St., Cambridge, MA 02138, e-mail: cushman@wjh. harvard.edu.

P S Y C H O L O G I C A L S C I E N C E

1082 Volume 17—Number 12Copyright r 2006 Association for Psychological Science

We investigated three principles:

� The action principle: Harm caused by action is morally worse than equivalent harm caused by omission.

� The intention principle: Harm intended as the means to a goal is morally worse than equivalent harm foreseen as the side

effect of a goal.

� The contact principle: Using physical contact to cause harm to a victim is morally worse than causing equivalent harm to a

victim without using physical contact.

The action principle has been well researched in psychology,

where it is often called omission bias (Baron & Ritov, 2004;

Spranca, Minsk, & Baron, 1991). The relevance of the action

principle is also recognized in the philosophical literature

(Quinn, 1989; Rachels, 1975). The intention principle, often

identified as the doctrine of the double effect, has received in-

tense scrutiny by philosophers (Foot, 1967; Nagel, 1986), but

markedly less by psychologists (but see Mikhail, 2002; Royz-

man & Baron, 2002). The contact principle has been com-

paratively understudied in both psychology and philosophy;

although it bears a superficial similarity to Greene’s distinction

between personal and impersonal moral dilemmas (Greene,

Sommerville, Nystrom, Darley, & Cohen, 2001), physical con-

tact is neither a necessary nor a sufficient condition for a per-

sonal moral dilemma.

Having established that subjects make use of a principle, one

can then ask whether this principle is available to conscious re-

flection during justification. On the one hand, we hypothesized

that a hallmark of conscious reasoning is that the principles used

in judgments are articulated in justifications. On the other hand,

we hypothesized that intuitive responses are accompanied by

insufficient justifications, uncertainty about how to justify, denial

of a relevant principle, or confabulation of alternative expla-

nations for judgments. Although it is possible that moral principles

consistently cited during justification were nonetheless engaged

without conscious reasoning during judgment, one may con-

clude that these principles are at least available for conscious

processes of moral reasoning. By contrast, those principles that

consistently cannot be cited during justification appear to be

genuinely inaccessible to conscious processes of reasoning.

METHOD

Subjects voluntarily logged on to the Moral Sense Test Web site,

moral.wjh.harvard.edu. Previous work with a different set of

dilemmas revealed no substantive differences in responses ob-

tained from subjects who answered questions on the Web and

those who completed more traditional pen-and-paper tests

(Hauser, Cushman, & Young, in press). Subjects were 37 years

old on average, and the sample had a small male bias (58%). We

instructed subjects to participate only if they were fluent in

English; 88% listed English as their primary language. Most

subjects indicated they were from the United States, Canada, or

the United Kingdom; 25% had been exposed to formal education

in moral philosophy.

After completing a demographic questionnaire, subjects re-

ceived 32 moral scenarios separated into two blocks of 16. Each

block included 15 test scenarios and 1 control scenario. Order of

presentation was counterbalanced between subjects, varying

both within and between blocks. For each scenario, subjects

rated the protagonist’s harmful action or omission on a scale from

1 to 7, with 1 labeled ‘‘Forbidden,’’ 4 labeled ‘‘Permissible,’’ and

7 labeled ‘‘Obligatory.’’ In a third block, subjects were asked to

justify their pattern of responses for up to five pairs of scenarios.

We asked subjects to justify only responses conforming to the

three principles being tested (e.g., when an action was judged

worse than an omission). Subjects were presented with the text of

the two scenarios side by side, reminded which they judged more

permissible, and asked to justify their pattern of responses.

All subjects had the opportunity to exit the testing session

after any number of blocks. We analyzed data only from subjects

who successfully completed all three blocks. Subjects were

omitted from all analyses if they failed either of the two control

scenarios (by judging permissible the killing or allowed death of

five people despite a costless alternative), or if they completed

any of the 32 scenarios in fewer than 4 s, deemed the minimum

comprehension and response time on the basis of pilot research.

Additionally, subjects were removed from the analyses of jus-

tifications if they misunderstood the task, provided a nonsen-

sical response, or provided a judgment that made it clear they

had misunderstood a scenario. These subjects were not removed

from our judgment analyses because not every subject justified

each judgment, thereby precluding the uniform application of

this procedure. Of 591 justifications, 65 were removed from the

analyses of justifications.

The test scenarios comprised 18 controlled pairs. What fol-

lows are brief descriptions of 4 scenarios; the actual text of all

32 (test and control) scenarios is available on the Web at moral.

wjh.harvard.edu/methods.html.

‘‘Evan’’ (action, intended harm, no contact): Is it permissible for Evan

to pull a lever that drops a man off a footbridge and in front of a

moving boxcar in order to cause the man to fall and be hit by the

boxcar, thereby slowing it and saving five people ahead on the tracks?

‘‘Jeff’’ (omission, intended harm, no contact): Is it permissible for

Jeff not to pull a lever that would prevent a man from dropping off a

footbridge and in front of a moving boxcar in order to allow the man

to fall and be hit by the boxcar, thereby slowing it and saving five

people ahead on the tracks?

‘‘Frank’’ (action, intended harm, contact): Is it permissible for

Frank to push a man off a footbridge and in front of a moving boxcar

in order to cause the man to fall and be hit by the boxcar, thereby

slowing it and saving five people ahead on the tracks?

‘‘Dennis’’ (action, foreseen harm as side effect, no contact): Is it

permissible for Dennis to pull a lever that redirects a moving

Volume 17—Number 12 1083

Fiery Cushman, Liane Young, and Marc Hauser

boxcar onto a side track in order to save five people ahead on the

main track if, as a side effect, pulling the lever drops a man off a

footbridge and in front of the boxcar on the side track, where he

will be hit?

Some scenarios belonged to more than one pair; for instance,

‘‘Evan’’ was contrasted with ‘‘Jeff’’ to yield an action-principle

comparison, with ‘‘Frank’’ to yield a contact-principle com-

parison, and with ‘‘Dennis’’ to yield an intention-principle

comparison (Fig. 1). Six pairs varied across the action principle,

six varied across the intention principle, and six varied across

the contact principle.

The methods used were in accordance with the regulations of

the institutional review board at Harvard University.

RESULTS

Judgments

Paired-sample t tests were performed on each of the 18 con-

trolled pairs of scenarios to determine whether subjects rated

one scenario in the pair significantly more permissible than the

other in the direction predicted by the relevant principle. Sta-

tistical significance was achieved in 17 out of 18 pairs at .05,

two-tailed (N 5 332); in the remaining pair, mean permissibility

ratings trended in the appropriate direction but fell short of

significance, p 5 .144 (Table 1). Across scenarios with different

content, subjects judged action as worse than omission, in-

tended harm as worse than foreseen harm, and harm via contact

as worse than harm without contact.

Justifications

A total of 526 justifications were coded for five nonexclusive

attributes. The attributes were as follows:

� Sufficiency: The subject mentioned a factual difference be- tween the two cases and either claimed or implied that it was

the basis of his or her judgments. It was not necessary for the

subject to identify the target principle, so long as the principle

generated by the subject could adequately account for his or

her pattern of responses on the scenario pair in question.

� Failure: The subject suggested an alternative principle, but this alternative could not account for his or her actual pattern

of judgments.

Fig. 1. Principle-based contrasts for four scenarios arranged into three pairs.

TABLE 1

Differences in Permissibility for Pairs of Moral Scenarios

Scenario pair Mean difference SD t(331) Effect size (d) p (two-tailed) prep

Action-principle pairs

Boxcar 0.70 2.03 6.32 0.34 <.001 >.999 Pond 1.69 2.00 15.34 0.84 <.001 >.999 Ship 0.83 2.01 7.56 0.41 <.001 >.999 Car 0.90 1.77 9.26 0.50 <.001 >.999 Boat 0.98 1.98 8.97 0.49 <.001 >.999 Switch 0.26 1.87 2.56 0.13 <.01 >.94

Intention-principle pairs

Speedboat 0.29 1.15 4.65 0.25 <.001 >.999 Burning 1.12 1.58 12.90 0.70 <.001 >.999 Boxcar 0.50 1.68 5.38 0.29 <.001 >.999 Switch 0.28 1.77 2.92 0.15 <.005 >.96 Chemical 0.24 1.51 2.91 0.15 <.005 >.96 Shark 0.30 1.77 3.14 0.16 <.003 >.978

Contact-principle pairs

Speedboat 0.89 1.44 11.36 0.62 <.001 >.999 Intended burning 0.24 1.40 3.18 0.17 <.003 >.978 Boxcar 1.07 1.72 11.28 0.62 <.001 >.999 Foreseen burning 0.37 1.22 5.50 0.30 <.001 >.999 Aquarium 0.17 1.35 2.31 0.12 <.022 >.947 Rubble 0.10 1.27 1.47 0.07 <.145 >.77

Note. All t tests were within subjects. Probability of replication (prep) was calculated according to Killeen (2005).

1084 Volume 17—Number 12

Conscious Reasoning and Intuition in Moral Judgment

� Uncertainty: The subject explicitly referenced his or her own uncertainty about how to justify the scenario, or directly

stated that he or she could not justify the responses.

� Denial: The subject stated that, at the point of justification, he or she did not consider there to be any moral difference be-

tween the two scenarios.

� Alternative explanation: The subject appealed to an alterna- tive explanation of his or her responses, either (a) invoking

facts that were not present in the scenarios or (b) claiming that

he or she made a mistake in selecting the appropriate re-

sponse, for instance, by clicking on the wrong button.

Our complete coding criteria, including examples from the data,

are available on the Web at moral.wjh.harvard.edu/meth-

ods.html. Justifications were coded by one of the authors (F.C.)

and by a colleague familiar with the research, and ties were

broken by another author (L.Y.). Table 2 provides the overall

percentage agreement between coders for individual attributes,

along with Cohen’s kappa, a statistic of interobserver reliability

for which values between .60 and .70 are considered fair, from

.70 to .75 are considered good, and above .75 are considered

excellent (Fleiss, 1981). For one of the attributes, failure, the low

Cohen’s kappa of .32 warrants caution in the interpretation of

results, although the overall agreement, 89%, was quite high.

Differences between principles in the proportion of justifi-

cations meeting criteria for each attribute were tested by chi-

square analysis (Table 2). For all five attributes, the proportion of

justifications meeting criteria differed significantly between

principles. Critically, subjects readily provided sufficient jus-

tifications for the action principle, rarely did so for the intention

principle, and showed an intermediate level for the contact

principle. Although justifications were coded as sufficient even

when subjects provided logically adequate justifications for

their behavior other than the target principle in question, we

observed that 95% of sufficient justifications depended on the

target principle. Subjects were more likely to provide failed

justifications or to express uncertainty for intention-principle

cases than for action-principle and contact-principle cases, and

were more likely to deny the moral relevance of the distinction or

to appeal to alternative explanations for intention-principle and

contact-principle cases than for action-principle cases.

It is possible that the attribute ‘‘alternative explanation’’

represents a baseline error rate for the task of judging pairs of

moral dilemmas; that is, subjects’ alternative explanations might

actually be true accounts of their behavior. If this interpretation

is accurate, then subjects should have been equally likely to make

an error in judgment on any trial, regardless of the principle that

the scenario was used to target during justification. Therefore,

the proportion of the total number of pairs of judgments made

that resulted in an alternative explanation during justification

should have been equal across the three principles. (In the

analyses described in the previous paragraph, the proportion

of justifications that relied on alternative explanations was

calculated over the total number of justifications of paired

scenarios, whereas in the analyses reported here, it was calcu-

lated over the total number of judgments of paired scenarios.)

Analyses revealed statistically significant differences among

these proportions, w2(2, N 5 1,502) 5 10.163, p 5 .006. Sub- jects were almost twice as likely to arrive at an alternative ex-

planation after judging a pair of intention-principle (9%) or

contact-principle (9%) scenarios than after judging a pair of

action-principle scenarios (5%). Alternative explanations

therefore seem to be driven by the principle demanded in jus-

tification, and do not exclusively represent a baseline error rate

for the task of judging moral dilemmas.

For a more stringent test of the differences in the attributes

elicited by the three principles, we treated controlled pairs of

scenarios, rather than individual justifications, as the unit of

analysis. After calculating for each scenario pair the proportion

of justifications meeting criteria for a given attribute, we used an

analysis of variance (ANOVA) to determine whether the mean

proportion for each attribute differed by principle. Though this

TABLE 2

Proportion of Justifications Exhibiting Each Attribute and Differences in Proportions Across Principles, With

Trials as the Unit of Analysis

Attribute

Proportion

Chi-square analysis

Interobserver reliability

Action- principle

pairs

Intention- principle

pairs

Contact- principle

pairs w2(2, N 5 526) p Cramer’s f

Overall percentage agreement Kappa

Sufficiency .81 .32 .60 91.149 <.001 0.42 89 .77 Failure .06 .16 .10 10.869 .004 0.14 89 .32

Uncertainty .05 .22 .04 39.058 <.001 0.27 97 .78 Denial .02 .17 .13 28.781 <.001 0.23 95 .67 Alternative explanation .10 .29 .32 34.344 <.001 0.26 90 .69

Note. For this analysis, the proportion of justifications exhibiting each attribute was calculated as the proportion of the total number of justifications provided for each principle. For each attribute, the chi-square analysis tested whether the proportions differed significantly by principle.

Volume 17—Number 12 1085

Fiery Cushman, Liane Young, and Marc Hauser

analysis reduced the sample size from 526 justifications to 18

scenario pairs, each ANOVA confirmed the previous set of

analyses (Table 3). To rule out the hypothesis that differences

between principles were driven by the size of the judgment

difference between items in a scenario pair, we repeated the

ANOVAs with mean judgment difference as a covariate. All five

ANOVAs still yielded significant differences by principle.

DISCUSSION

This study presents evidence for three principles that guide

judgments of moral dilemmas, each focused on trade-offs of life

and death, but varying in content. Subjects rated harmful actions

as morally worse than harmful omissions (the action principle),

harm intended as the means to an end as morally worse than

harm foreseen as the side effect of an end (the intention prin-

ciple), and harm involving physical contact as morally worse

than harm without contact (the contact principle).

The content of subjects’ justifications differed greatly by

principle. In the case of the action principle, a large majority of

subjects were able to provide sufficient justifications for their

judgments, whereas relatively few provided failed justifications,

denied any moral difference between the scenarios, or expressly

doubted their ability to justify their responses. These data are

consistent with the conscious-reasoning model of moral judg-

ment. Although it is possible that subjects constructed the ac-

tion principle post hoc upon having to justify their responses, a

large majority of subjects at least possessed the requisite ex-

plicit knowledge required by the conscious-reasoning account.

Quite the opposite is true for the intention principle. Less than

a third of subjects were able to provide sufficient justifications

for their pattern of judgments regarding the intention-principle

pairs. In 22% of the justifications, subjects specifically indi-

cated uncertainty about how to justify their responses; in 17% of

the justifications, subjects denied that there was any morally

relevant difference between the cases; and 16% of the justifi-

cations in fact failed to account for the subject’s pattern of

judgments. These data are more consistent with an intuitionist

model of moral judgment. Subjects reliably generated a pattern

of moral judgments consistent with the intention principle, but

were generally incapable of articulating it, to the point of ex-

pressing confusion when confronted with their own judgments or

even denying their judgments altogether. The intention prin-

ciple clearly plays a role in moral judgment, but most likely in

the form of unconscious—or at least inexpressible—knowledge.

It is notable that subjects were almost three times as likely to

invoke alternative explanations for their responses to intention-

principle cases as to invoke alternative explanations for their

responses to action-principle cases, either by claiming to have

made an error (e.g., by pushing the wrong button) or by invoking

unwarranted assumptions about the scenarios (e.g., assuming

that a harm would actually not occur, even when the scenario

explicitly stated that it would). There is no reason to believe that

the particular scenarios used in intention-principle pairs were

any more likely to invite either errors or assumptions than those

used in action- or contact-principle pairs—the scenarios were

identically constructed, and indeed, in several cases individual

scenarios were used in contrasts of both types. Thus, what de-

termines the use of alternative explanations appears not to be the

scenario presented during judgment, but rather the principle

targeted in justification. We conclude that in some cases, sub-

jects’ appeals to alternative explanations were confabulations

generated at the moment of justification and prompted by the

inability to justify the pattern of judgments. Parallel evidence

comes from Wheatley and Haidt’s (2005) recent study of priming

under hypnosis: Subjects’ confabulation accompanied their in-

ability to provide a principled justification of moral judgment.

Subjects’ justifications of their responses to contact-principle

cases occupied an intermediate position between justifications

for action-principle and intention-principle cases. Subjects

were typically able to articulate the relevant principle used, but

were relatively unwilling to endorse it as morally valid. Sixty

TABLE 3

Proportion of Justifications Exhibiting Each Attribute and Differences in Proportions Across

Principles, With Scenario Pairs as the Unit of Analysis

Attribute

Mean proportion

F(2, 15) p Zp 2

Action-principle pairs

Intention-principle pairs

Contact-principle pairs

Sufficiency .79 .30 .59 22.44 <.001 .749 Failure .07 .18 .09 5.75 .014 .434

Uncertainty .05 .23 .04 12.28 .001 .621

Denial .02 .20 .14 8.14 .004 .520

Alternative explanation .11 .29 .34 8.60 .003 .534

Note. For this analysis, the proportion of justifications exhibiting each attribute was calculated separately for each scenario pair. The mean proportions of the six scenario pairs that tested each principle are listed under ‘‘Mean pro- portion.’’ The analysis of variance tested whether the mean proportions differed by principle.

1086 Volume 17—Number 12

Conscious Reasoning and Intuition in Moral Judgment

percent of justifications in contact-principle cases were suffi-

cient, nearly twice the proportion for intention-principle cases.

Additionally, only 10% of justifications were classified as failed,

and 4% indicated uncertainty. These statistics are all compa-

rable to those for action-principle cases. Yet in 13% of contact-

principle justifications, subjects denied that physical contact

made a moral difference, a proportion much closer to that for the

intention-principle cases (17%) than for the action-principle

cases (2%). Subjects were also about three times more likely to

appeal to alternative explanations in contact-principle cases

than in action-principle cases—a pattern of results nearly

identical to that for the intention principle. Thus, although

subjects were able to articulate the principle behind their rea-

soning in contact-principle cases, they often rejected it as

morally invalid, or confabulated alternative explanations for

their behavior. For instance, one subject wrote, ‘‘I guess I was

fooled by the line-pulling seeming more passive than the man-

pushing, but that view is hard to justify now.’’

Although a conscious-reasoning interpretation of subjects’

justifications for contact-principle cases cannot be rejected

definitively, the data favor the intuitionist view. The observation

that many subjects used, but were unwilling to endorse, the

contact principle is not readily explained by the conscious-

reasoning model. Why would a subject reason consciously from

an explicit principle about physical contact during judgment,

but then disavow the same principle during justification? A more

plausible explanation is that the contact principle guides moral

judgments according to the intuitionist model during judgment,

and that a process of post hoc reasoning at justification allows

subjects to deduce the principle behind their judgments. Once

deduced, the principle is regarded as morally irrelevant. Just

such a process of post hoc reasoning has been proposed by Haidt

(2001). This interpretation of contact-principle cases raises the

question of why a similar process was not observed for the in-

tention-principle cases. One possible explanation is that the

factors over which the intention principle operates are less sa-

lient than those over which the contact principle operates; this

speculation requires additional research.

In summary, our results show that although some moral

principles are available for conscious reasoning in a large ma-

jority of subjects, others are not available and appear to operate

in intuitive processes. The extent to which conscious reasoning

or intuition plays a more dominant role depends on the par-

ticular moral principles triggered: The intention principle seems

best characterized by the intuitionist model, the contact prin-

ciple by intuitive judgment followed by rational reflection, and

the action principle either by conscious reasoning or, at a

minimum, by post hoc conscious reasoning.

The present findings bear in important ways on issues central

to the study of moral cognition. We briefly raise three such is-

sues. First, our results underscore the methodological impor-

tance of linking subjects’ expressed moral principles to the

operative principles underlying their patterns of judgment

(Hauser, in press; Mikhail, 2000). Controlled pairs of scenarios

have long been used in philosophy and psychology to probe

moral judgments (Kamm, 2001; Petrinovich, O’Neill, & Jor-

gensen, 1993; Spranca et al., 1991; Thompson, 1985), but have

not been extended to moral justification. Without using tightly

controlled scenario pairs to target narrowly defined principles, it

is not possible to determine whether a subject’s justification can

account for his or her judgments. It has been demonstrated that

under uniform testing conditions, this method can yield data that

distinguish principles available to conscious reasoning from

those that are not. Having demonstrated divergent results under

a common paradigm, researchers are better equipped to detect

genuine differences in the application of moral principles.

This methodological point has been underappreciated by

advocates of both conscious-reasoning and intuitionist models.

On the one hand, within the tradition established by Kohlberg

(1969), the moral principles cited in justification were simply

assumed to be identical to the moral principles used in judg-

ment. Our data showing a high rate of failed justifications for

intention-principle cases suggest that this assumption is not

always valid. On the other hand, Haidt (2001) demonstrated

moral dumbfounding (the inability to justify a moral judgment)

by asking subjects why, for instance, it is wrong for a brother and

sister to have intercourse, but did not demonstrate that subjects’

patterns of judgments necessitate a principle more complex than

‘‘it is wrong for a brother and a sister to have intercourse.’’ In the

paradigm we used in the present study, dumbfounding might

arise if subjects were asked why a harmful action is worse than a

harmful omission. But this is an altogether different question

from whether subjects have explicit knowledge of the principle

to which their pattern of judgments conforms, which is simply

that a harmful act is less permissible than a harmful omission.

From either a conscious-reasoning or an intuitionist perspec-

tive, a proper analysis of subjects’ justifications cannot be

conducted in the absence of a precise accounting of their

judgments. However, because of the possibility of post hoc

reasoning in cases in which subjects provide sufficient justifi-

cations, the present method is better suited to rejecting the role

of conscious reasoning than supporting it. A task for future

studies is to design methodologies that provide strong evidence

in favor of consciously reasoned moral judgments.

Second, our results provide evidence for moral principles that

are formulated over detailed representations of the causal and

intentional aspects of harm. To date, many of the moral principles

proposed by psychologists have far broader content than those we

have presented. Kohlberg’s (1969) moral stages were formulated

over concepts like authority, cooperation, and autonomy. Greene

has proposed an emotion-based appraisal system with a general

prohibition of actions fitting the semantic structure ‘‘me hurt you’’

(Greene, Nystrom, Engell, Darley, & Cohen, 2004). Haidt has

formulated moral principles over the concepts of suffering, reci-

procity, hierarchy, purity, and group identity (Haidt, in press).

Ultimately, these useful generalizations will have to be married to

Volume 17—Number 12 1087

Fiery Cushman, Liane Young, and Marc Hauser

a more detailed analysis of basic concepts such as action, in-

tention, and contact (Hauser, Cushman, & Young, in press;

Hauser, Young, & Cushman, in press; Knobe, 2003; Mikhail,

Sorrentino, & Spelke, 2002; Pizarro et al., 2003).

Finally, our results support the view that moral judgment can

be accomplished by multiple systems: Some moral principles

are available to conscious reflection—permitting but not guar-

anteeing a role for conscious reasoning—whereas others are

better characterized by an intuitionist model. On the basis of

neuroimaging data, Greene et al. (2004) have proposed a two-

system model for moral judgment in which one system is char-

acterized by the engagement of affective systems and the other is

characterized by the engagement of cognitive systems. The current

findings suggest that, regardless of where the division between

affect and cognition is placed, a multisystem model of moral

judgment is warranted. Such a multisystem model stands in con-

trast to Kohlberg’s (1969) perspective, in which all moral rea-

soning is assumed to be the product of conscious reasoning. It also

stands in contrast to the recent intuitionist proposal by Mikhail

(Mikhail et al., 2002) to incorporate the sorts of moral principles

discussed here into a single evaluative mechanism that accom-

plishes moral judgment in a manner loosely analogous to the way

in which syntactic structure is analyzed in language.

In conclusion, this article has outlined a novel methodological

approach to the study of moral psychology, highlighting the

interaction of intuition and conscious reasoning and empha-

sizing the distinction between the principles that people use and

the principles that people articulate.

Acknowledgments—We thank Jennifer Pegg for her contri-

bution to the analysis of the justification data, and Jonathan

Baron, Jonathan Haidt, David Pizarro, and Joshua Knobe for

their comments on an earlier version of this manuscript.

REFERENCES

Bargh, J.A. (1999). The unbearable automaticity of being. American Psychologist, 54, 462–479.

Baron, J., & Ritov, I. (2004). Omission bias, individual differences,

and normality. Organizational Behavior and Human Decision Processes, 94, 74–85.

Fleiss, J.L. (1981). Statistical methods for rates and proportions (2nd ed.). New York: Wiley.

Foot, P. (1967). The problem of abortion and the doctrine of double

effect. Oxford Review, 5, 5–15. Greene, J.D. (in press). The secret joke of Kant’s soul. In W. Sinnott-

Armstrong (Ed.), Moral psychology: Vol. 3. The neuroscience of morality. Cambridge, MA: MIT Press.

Greene, J.D., & Haidt, J. (2002). How (and where) does moral judg-

ment work? Trends in Cognitive Sciences, 6, 517–523. Greene, J.D., Nystrom, L.E., Engell, A.D., Darley, J.M., & Cohen, J.D.

(2004). The neural bases of cognitive conflict and control in moral

judgment. Neuron, 44, 389–400.

Greene, J.D., Sommerville, R.B., Nystrom, L.E., Darley, J.M., & Co-

hen, J.D. (2001). An fMRI investigation of emotional engagement

in moral judgment. Science, 293, 2105–2108. Haidt, J. (2001). The emotional dog and its rational tail: A social in-

tuitionist approach to moral judgment. Psychological Review, 108, 814–834.

Haidt, J. (in press). Social intuitionists answer six questions about

moral psychology. In W. Sinnott-Armstrong (Ed.), Moral psychol- ogy: Vol. 2. The cognitive science of morality. Cambridge, MA: MIT Press.

Haidt, J., & Hersh, M.A. (2001). Sexual morality: The cultures and

emotions of conservatives and liberals. Journal of Applied Social Psychology, 31, 191–221.

Hauser, M.D. (in press). Moral minds: How nature designed a universal sense of right and wrong. New York: Harper Collins.

Hauser, M.D., Cushman, F.A., & Young, L. (in press). A disso-

ciation between moral judgment and justification. Mind and Language.

Hauser, M.D., Young, L., & Cushman, F.A. (in press). Reviving Rawls’

linguistic analogy. In W. Sinnott-Armstrong (Ed.), Moral psychol- ogy: Vol. 2. The cognitive science of morality. Cambridge, MA: MIT Press.

Kamm, F.M. (2001). Morality, mortality: Rights, duties, and status. New York: Oxford University Press.

Killeen, P.R. (2005). An alternative to null-hypothesis significance

tests. Psychological Science, 16, 345–353. Knobe, J. (2003). Intentional action in folk psychology: An experi-

mental investigation. Philosophical Psychology, 16, 309–324. Kohlberg, L. (1969). Stage and sequence: The cognitive-develop-

mental approach to socialization. In D.A. Goslin (Ed.), Handbook of socialization theory and research (pp. 151–235). New York: Academic Press.

Mikhail, J. (2002). Aspects of the theory of moral cognition: Investi- gating intuitive knowledge of the prohibition of intentional battery and the principle of double effect (Georgetown Public Law Re- search Paper 762385). Washington, DC: Georgetown University

Law Center.

Mikhail, J.M. (2000). Rawls’ linguistic analogy: A study of the ‘gen- erative grammar’ model of moral theory described by John Rawls in ‘A theory of justice,’ Unpublished doctoral dissertation, Cornell University, Ithaca, NY.

Mikhail, J.M., Sorrentino, C., & Spelke, E. (2002). Aspects of the theory of moral cognition: Investigating intuitive knowledge of the prohibition of intentional battery, the rescue principle, the first principle of practical reason, and the principle of double effect. Unpublished manuscript, Stanford University, Stanford, CA.

Nagel, T. (1986). The view from nowhere. Oxford, England: Oxford University Press.

Petrinovich, L., O’Neill, P., & Jorgensen, M.J. (1993). An empirical

study of moral intuitions: Towards an evolutionary ethics. Journal of Personality and Social Psychology, 64, 467–478.

Pizarro, D.A., & Bloom, P. (2003). The intelligence of the moral in-

tuitions: Comment on Haidt (2001). Psychological Review, 110, 193–198.

Pizarro, D.A., Uhlmann, E., & Bloom, P. (2003). Causal deviance and

the attribution of moral responsibility. Journal of Experimental Social Psychology, 39, 653–660.

Quinn, W.S. (1989). Actions, intentions, and consequences: The

doctrine of doing and allowing. The Philosophical Review, 145, 287–312.

1088 Volume 17—Number 12

Conscious Reasoning and Intuition in Moral Judgment

Rachels, J. (1975). Active and passive euthanasia. New England Journal of Medicine, 292, 78–80.

Royzman, E., & Baron, J. (2002). The preference for indirect harm.

Social Justice Research, 15, 165–184. Spranca, M., Minsk, E., & Baron, J. (1991). Omission and commission

in judgment and choice. Journal of Experimental Social Psy- chology, 27, 76–105.

Thompson, J.J. (1985). The trolley problem. The Yale Law Journal, 94, 1395–1415.

Wheatley, T., & Haidt, J. (2005). Hypnotic disgust makes moral

judgments more severe. Psychological Science, 16, 780–784.

(RECEIVED 11/21/05; REVISION ACCEPTED 2/15/06; FINAL MATERIALS RECEIVED 3/10/06)

Volume 17—Number 12 1089

Fiery Cushman, Liane Young, and Marc Hauser