Morality Influences How People Apply the Ignorance of the Law Defense

Abstract

In four empirical studies, we showed that laypeople apply the ignorance of the law defense differently depending on the perceived morality of the defendant's course of conduct at the time of the illegal act. Moral and neutral defendants who pled ignorance of the law were afforded leniency, whereas immoral defendants were sentenced as though they were not ignorant, even when defendants in all three conditions violated identical laws. These findings suggest that laypeople adopt a just deserts approach to criminal law, which influences their responsiveness to a criminal defendant's claim to be ignorant of the law. We discuss the implications of these findings for criminal law and argue that legal doctrine should reflect laypeople's moral intuitions.

The contention that an injury can amount to a crime only when inflicted by intention is no provincial or transient notion. It is as universal and persistent in mature systems of law as belief in freedom of the human will and a consequent ability and duty of the normal individual to choose between good and evil. A relation between some mental element and punishment for a harmful act is almost as instinctive as the child's familiar exculpatory “But I didn't mean to[.]”

(Morissette v. United States, 342 U.S. 246, 250–51 [1952])

Criminal conviction has two prerequisites: the actus reus (criminal act) combined with mens rea (criminal intent). The mens rea requirement is a hurdle intended to ensure that the criminal law punishes only the blameworthy or culpable, those individuals who consciously choose to do wrong. Consequently, reasonable mistakes of fact and mistakes of collateral (nonpenal) law generally provide a defense to conviction when the mistake negates the mental state element of the crime. Yet since American law's earliest roots, ignorance or mistake of penal law (the law that criminalizes the act with which the defendant is charged) has not been a defense to criminal conviction (Davies 1998). This rule is sometimes referred to as the “ignorance maxim.”

Commentators have long recognized that the ignorance maxim is potentially at odds with the principle that criminal law punishes only the blameworthy (Holmes 1881). If a person is truly unaware that his or her conduct violates the law, then how can that person form the criminal intent necessary for conviction? This prospect of punishing the ignorant “stirs large questions—questions that go to the moral foundations of the criminal law” (United States v. Int'l Minerals & Chem. Corp. 1971, J. Stewart, dissenting). To overcome these “large questions,” many utilitarian justifications have been proffered for denying the ignorance of law defense: the difficulty of disproving a claim of ignorance, the goal of encouraging citizens to know and obey the law, and the need to prevent an accused person's idiosyncratic interpretation from attaining the status of a legal rule (Holmes 1881; Davies 1998:350–4). As jurist Oliver Wendell Holmes theorized, “It is no doubt true that there are many cases in which the criminal could not have known that he was breaking the law, but … justice to the individual is rightly outweighed by the larger interests on the other side of the scales” (Holmes 1881:48). These utilitarian justifications for denying the ignorance defense are persuasive as far as they go, but they ultimately fail to resolve the moral tension at the core of the maxim that “ignorance does not excuse.”

Criminal law is generally understood to codify society's reaction to acts that are morally blameworthy. In other words, criminal law differs from civil law in that it punishes rather than merely regulates. This article proceeds from the premise that in order to sustain its moral authority, the criminal law should strive as much as possible to remain in step with the basic moral intuitions of the surrounding community. We recognize that morality is a slippery and by no means unitary concept. Moral consensus is not always achievable in a diverse and free society, nor is majority rule always desirable. However, we do propose that conformity with community mores, to the extent that consistent mores can be identified, should be one of the chief guiding principles in the development of criminal rules and doctrine. It follows that moral intuitions can and should be measured. In that spirit, this article presents a group of four studies that provide the first empirical investigation of laypeople's (non-lawyers') moral intuitions about the interaction between the defense of legal ignorance and the defendant's broader course of conduct.

This assessment of moral intuitions about the ignorance maxim is especially urgent in light of the contemporary profusion of regulatory crimes. At common law, it was reasonable to assume that all competent adults were aware that crimes such as murder or arson were prohibited. These are classic examples of mala in se crimes, acts that will be recognized as wrong by any community member with a functioning moral compass. But in the era of the modern regulatory state, American criminal law has expanded far beyond this category and into a large realm of acts that are mala prohibita, wrong primarily because lawmakers have banned them. For example, California law authorizes criminal sanctions for landowners who allow the waste of water from artesian wells on their property (Cal. Water Code §307). Regulatory offenses in federal law and state codes include crimes of omission and can be punishable as felonies (Coffee 1991). In many instances, regulatory crimes are also established as “public welfare offenses” subject to strict liability, meaning that mens rea is not an element of the crime. Initially a narrow exception to the rule that criminal intent is required for conviction, public welfare offenses are proliferating. In the regulatory state of today, it is unrealistic to expect any citizen to be aware of all the criminal rules in effect, and it is increasingly plausible that an individual may commit a crime by doing an act that does not appear, according to his moral common sense, to be wrongful. As a result, the moral tension between the ignorance maxim and the claim of criminal law to punish only wrongful acts is steadily rising. Notwithstanding the confidence with which courts and professors may cite the maxim as a blanket rule, some legal decisions have responded to this tension by offering relief to individual defendants in cases where a rigid application of the ignorance maxim would create results that the courts judged intolerable.

The U.S. Supreme Court has never explicitly permitted the ignorance defense, but the Court has acknowledged and tried to ease this moral tension in several federal criminal appeals. In these cases, rather than squarely reconsidering the ignorance maxim, the Court has interpreted specific federal criminal statutes in order to reinvigorate mens rea by requiring knowledge of illegality as a prerequisite for conviction. In Liparota v. United States (1985), the Court found that a statute criminalizing “knowing” possession of food stamps required the government to prove not only that the defendant knew he possessed food stamps, but also that the defendant knew this possession was illegal. The Court reasoned that, absent clear legislative intent to depart from the usual rule that mens rea is required for conviction, legislation could not be interpreted so as to “criminalize a broad range of apparently innocent conduct” (Liparota v. United States 1985:426). Because possession of food stamps is an activity that appears innocent in light of commonsense notions of morality, it could not be assumed that knowing possession is equivalent to criminal intent.

Nine years after Liparota, the Court applied the same reasoning to find that awareness of illegality is required for conviction under the federal statute that prohibits the “willful” structuring of financial transactions to avoid federal reporting requirements (Ratzlaf v. United States 1994). Ratzlaf, a gambler, paid a large casino debt with a series of cashier's checks in amounts just below the reporting threshold, shortly after the casino warned him of its obligation to report transactions over $10,000 to the government. Although Ratzlaf's behavior showed a clear intent to evade the reporting law, the Court ruled that, because the evasive tactic of financial structuring is not itself “obviously evil,” (Ratzlaf v. United States 1994:73) the government could not obtain a conviction without proving that Ratzlaf knew financial structuring to be illegal. In both Liparota and Ratzlaf, the Court concluded that a defendant of competent morality might nevertheless have reasonably not suspected that his actions were regulated by law. Under these circumstances, the Court found that criminal liability was inappropriate.

The Court went somewhat farther in Cheek v. United States (1991), which reversed the conviction of an airline pilot who was convicted of willful tax evasion after repeatedly failing to file his income tax returns. The Supreme Court ruled that Cheek should have been allowed to present evidence to the jury showing that he honestly believed that wages are not income, even though such a belief was “not objectively reasonable” (Cheek v. United States 1991:192). Earlier cases had permitted a reasonable mistake defense in tax prosecutions on the rationale that tax laws are so complex that the average citizen may violate them despite good faith efforts to comply (United States v. Murdock 1933). However, allowing Cheek to claim a defense of unreasonable mistake of law represented a significant expansion of this tax exception and a step beyond the earlier case of Liparota, which had emphasized the apparent innocence of the regulated behavior. In Cheek, the Court placed a very high importance indeed on the goal of protecting the morally blameless from criminal sanction.

The Supreme Court's cases on ignorance of the law are similar to another set of decisions in which the Court has interpreted statutes to include mistake of fact defenses in order to avoid criminalizing innocent conduct (Staples v. United States 1994; United States v. X-Citement Video 1994). In the mistake of fact cases, the Supreme Court likewise emphasized the foreseeability of regulation and—again, despite unsympathetic defendants—concluded that both guns and adult magazines were not so widely regulated as to put owners and distributors on notice that their conduct might be subject to criminal sanctions.

The Supreme Court's precedents on ignorance of law and fact, though scattered and narrowly written, are linked by a common philosophy and have been hailed by previous commentators as an emerging jurisprudence of “mandatory culpability” (Wiley 1999:1023) or “excusable ignorance” (Davies 1998:341). In all of these cases, the Court emphasized the theme of foreseeability or notice, reasoning that if the underlying illegal act is not obviously wrongful, then criminal punishment is not appropriate unless the defendant acted wrongfully by intentionally disobeying the law. The Court rooted these decisions, sometimes explicitly, in a concern that the apparent innocence of the regulated conduct permits prosecutors to unleash the criminal law on unwitting and therefore blameless defendants (Davies 1998:377). In part, the Court employed this reasoning to limit the proliferation of strict liability crimes and to reinvigorate the concept of mens rea. At the same time, the Court went farther and betrayed a certain discomfort over the universal application of the ignorance maxim. In instances where enforcing the maxim had a plausible chance of convicting those whose acts were illegal but perhaps not a symptom of moral failure, the Court responded to this discomfort by reading an ignorance of the law defense into isolated federal statutes in order to provide a safety valve for defendants who broke the law unwittingly. The Court openly invoked notions of commonsense morality in the ignorance cases, and the language of these opinions—heavily reliant on terms such as “innocent” and “evil”—is the language of moral conflict. The Court's approach in these cases, especially in the extreme form seen in the Cheek case, illustrated an intent-based approach to criminal law, in which an actor's state of mind is more relevant than the results of the action.

Nearly half a century ago, Hart argued that the ban on the ignorance defense is based on an understanding that a person who engages in inherently wrongful acts is culpable for failing to know about and abide by society's moral consensus (H. M. Hart 1958). Yet it is only fair to assume that a person who is ignorant of the law is also ignorant of morality as long as the law and morality are essentially consonant. Hart therefore warned that the maxim should only apply in cases of mala in se crimes where there is societal consensus about the wrongfulness of the prohibited behavior. Hart's admonition highlights the importance of societal norms for resolving the dilemma of the ignorance defense. In a more recent analysis, Kahan (1997) reached similar conclusions about the moral dynamic of cases where ignorance of the law is raised. Kahan claimed that the traditional utilitarian justifications for disallowing the ignorance defense do not sufficiently explain the outcomes of cases. Supporting his theory with illustrative case studies, Kahan argued that courts accept the ignorance defense selectively in cases where the mistake of law asserted by the defendant does not mark the boundary between moral and immoral behavior. In other words, courts generally do allow ignorance defenses for malum prohibitum crimes. When a court does not wish to allow the defense, it can recite the ignorance maxim and disallow the defense. Kahan framed his argument in terms of two primary categories of defendants, the immoral and the virtuous. However, he did not probe the gray region between morality and immorality, which could be described as morally neutral.

Studies of Laypeople's Views on Ignorance of the Law

Recent research has uncovered important reasons to take seriously laypeople's attitudes toward the legal system. The criminal law is designed to maintain order, but it cannot eliminate lawlessness if people believe it is either ineffective or unworthy of respect (e.g., Carlsmith et al. 2002; Darley et al. 2000). To this effect, Nadler (2005; see also Greene 2003) recently demonstrated that people are more likely to violate minor laws when the legal system fails to reflect their conceptions of justice. In order to evaluate whether the legal system reflects people's conceptions of justice, it is important to conduct studies that expose laypeople's attitudes to various aspects of the legal system.

To that end, in this article we explore how citizens reason about ignorance of the law and about the morality of defendants' conduct. Faced with a malum prohibitum crime, laypeople may feel that it is fair for the court to entertain an ignorance defense. Alternatively, laypeople may be persuaded by the traditional utilitarian arguments in favor of denying the ignorance defense, regardless of the morality of the defendant's course of conduct. We suspected that laypeople share the ambivalence toward the ignorance defense that lawyers have long expressed through judicial decisions and legal commentary. We set out to test this empirically by investigating laypeople's responses to claims of legal ignorance raised by defendants who appeared moral, immoral, or morally neutral. As mentioned at the outset, our focus is on the interaction between the defense of legal ignorance and the defendant's broader course of conduct.

Consider two hypothetical cases in which defendants are charged for conduct that they mistakenly believe to be legal. The defendant in the first case is a coral reef researcher. In the course of her efforts to preserve endangered coral species, she collects substantial samples of dead coral in a way that she knows will not damage the fragile reef. She is unaware of a new law prohibiting any removal of dead or living coral but is arrested and tried for violating the law. Would you, as a juror, vote to convict?

In the second case, the defendant is the CEO of a cigarette manufacturing company. Believing that cigarette advertisements are banned only in hard copy form, he advertises cigarettes on several Internet sites that cater exclusively to an audience of teenagers, most of whom are too young to smoke legally. He is unaware of a new law extending the ban on cigarette advertising to the Internet but is arrested and tried for violating the law. Would you, as a juror, vote to convict?

Both the coral researcher and cigarette company CEO were ignorant of the laws they were accused of violating. To examine laypeople's intuitions about how the criminal law should respond to such claims of ignorance, we ran a series of studies in which we asked people to read similar mock cases, to decide whether the defendant should be convicted, and to suggest an appropriate sentence for each case. We do not claim that our results explain the outcomes of actual cases. We recognize that in common law systems, in which judges make legal determinations and juries make fact determinations, jurors would not actually choose whether to permit an ignorance defense or determine its effect on sentencing. Generally speaking, jurors would determine whether the defendant has proven ignorance (or the government fails to prove knowledge) only if the court permits an ignorance claim in the first instance and admits evidence relevant to the claim. Likewise, in the vast majority of criminal systems in the United States, juries do not make sentencing decisions other than in capital cases, which so clearly involve mala in se crimes that our thesis did not apply. Our studies also did not approximate the complex process of jury decision making. Rather, our studies measured laypeople's views as a means to gauge public moral intuitions about the ignorance defense. We asked respondents to provide sentencing and guilt determinations because these criminal law concepts are familiar to laypeople untrained in the law, and in order to capture the moral judgment that is associated with the acts of assigning criminal blame and imposing criminal punishment.

We predicted that people would react very differently to cases such as the two mentioned above. The U.S. Supreme Court, without explicitly overturning the ignorance maxim, has expressed unease with denying the ignorance defense in cases where a defendant may be morally blameless. Kahan (1997) has claimed that the same moral evaluation occurs in lower courts. We anticipated that the hypothetical cases above would not be widely perceived as moral close calls, and that most people would perceive saving coral reefs as a virtuous activity and advertising cigarettes to underage youth as a reprehensible activity. Consistent with courts' ambivalence toward the ignorance of law defense, we therefore predicted that, faced with equally plausible claims of ignorance from two defendants, laypeople would be inclined to acquit the morally virtuous coral researcher and convict the morally corrupt CEO. Simply, we believed that people are more likely to wish to permit ignorance of the law as a defense, or to perceive ignorance as a mitigating factor in sentencing, when the defendant's behavior was otherwise moral and well-intentioned.

Finally, we believed that people's moral intuitions would lead them to make predictable sentencing suggestions (see, e.g., Finkel 2001). Specifically, people will give defendants the benefit of the doubt if their ignorance of a law is plausible, unless the course of conduct surrounding the law-violating act is viewed as immoral. This prediction is consistent with recent empirical work that tends to show that the principle of “just deserts” is the primary motive that drives ordinary people's sentencing judgments (Carlsmith et al. 2002; see also Darley et al. 2000; H. L. A. Hart 1958; Packer 1968). Under a just deserts scheme, punishment severity is strongly determined by the moral wrongfulness of the offender's behavior. Thus minor crimes are punished leniently and serious felonies are punished more harshly. Other considerations, such as deterrence, do not have as great an effect on laypeople's sentencing preferences. Given the importance of just deserts in the criminal context, we would expect that a person motivated by just deserts considerations would not wish to punish a defendant who was doing a morally good act and was ignorant of the law prohibiting that act, but would want to punish a defendant who was engaged in an immoral course of conduct, even if ignorant of the illegality of his or her actions.

In sum, we predicted that respondents would be more likely to acquit moral actors who ignorantly violated laws or, if they did decide to convict those actors, would assign a comparatively lenient punishment. On the other hand, we predicted that people who committed morally wrong actions would be judged guilty more frequently and would receive harsher punishments for an equivalent offense, even if they were ignorant of the law banning their conduct. Put differently, we expected that laypeople, like courts and legal commentators, would be most likely to deny the ignorance defense to a defendant whose conduct would have been immoral even if the law had been as the defendant believed it to be, and would be most likely to afford the ignorance defense to a defendant whose conduct was illegal but otherwise moral.

We were less certain about how respondents would treat actors who ignorantly committed a morally neutral offense. For example, we believed that a person who unwittingly fishes in an area that requires a license is neither behaving morally nor immorally, and is thus behaving in a morally neutral manner. We expected that, in the absence of immorality, participants would substantially treat the defendant as they would a moral defendant. In other words, we believed that the default position when a person unwittingly commits a legal wrong would be for sentencers to adopt a lenient stance. This prediction indicates our view that, when responding to the moral challenge of the ignorance defense, laypeople (like courts) would be primarily motivated by a wish to avoid punishing the morally blameless, rather than a desire to reward the virtuous.

By using a within-subjects design in the first study, we presented each respondent with the direct question of whether he or she consciously thought there should be a distinction between the two cases. This is a powerful design that is likely to uncover weaker effects—such as differences in sentencing of moral and neutral defendants—as it allows participants to evaluate each case relative to the others. In the second, third, and fourth studies we used a between-subjects design to conduct a more conservative test of our hypotheses. In these studies, the respondents did not make the direct comparison of cases that would lead them to focus consciously on the roles of morality and legal ignorance in perceptions of crime severity.

Study 1

Method

Participants

Seventeen Princeton University psychology undergraduates (11 females) participated in this study to satisfy a course requirement. Participants were aged between 17 and 22 years (M=19.43 years, SD=1.28).

Materials, Design, and Procedure

The experimental stimuli were descriptions of offenders inadvertently committing crimes and being apprehended by the police (see Appendix A for full text of descriptions). We designed the experimental stimuli so that the moral offender behaved in a moral fashion, the immoral offender behaved immorally, and in the neutral stimulus, the actor behaved neither morally nor immorally. Specifically, the moral defendant was a coral researcher who, in the course of attempting to find a solution to coral bleaching, removed dead coral from a nature reserve and thereby violated an obscure law prohibiting the removal of coral, dead or alive, from the reef. The neutral defendant was an 18-year-old male student who lived in New Jersey and knew that the legal age of consent in New Jersey was 16 years of age. However, he met a 16-year-old girl and went to her home in New York City, where they had sexual intercourse. The girl's father, a lawyer, discovered this and, knowing that the legal age of consent was 17 years in New York, called the police, who charged the student with statutory rape. Finally, the immoral defendant was the head of a European drug syndicate. He entered the United States to buy drugs but, before he made the purchase, the police discovered that he was carrying a large sum of money. He was unaware of the customs rules that prevent visitors from entering the country with very large sums of money, and he was charged with a customs offense.

This study was a three-level, one-way, within-subjects experiment. We analyzed the data using a within-subjects analysis of variance (ANOVA). We decided to conduct this first study using a within-subjects design to encourage participants to focus on the distinctions between the cases in the different conditions. The experimental stimuli were presented in a counterbalanced order, and three filler stimuli were always presented second, fourth, and sixth. The filler stimuli were similar to the experimental stimuli, except that the defendants did not claim ignorance as a defense. As such, we hoped to prevent participants from detecting the purpose of our study.

Participants only responded to the stimuli once they had read all of them, so that they were able to form comparative judgments of the stimuli as a complete set. To minimize demand characteristics, we told participants only that they would read several scenarios and answer questions based on the scenarios. Accordingly, we did not tell them to attend to the targets' ignorance of the law, as we wanted them to respond to all of the information in each scenario and to use information about the participants' intent as a juror might do so naturalistically. (See Appendix B for a full list of instructions. The same instructions and questions generally applied to the other studies, except where exceptions are noted in the text.) All participants read the same scenarios: one experimental stimulus from each of three morality conditions (moral, neutral, immoral), and three filler scenarios in which the defendants were not ignorant of the laws that they violated. We included the filler stimuli to prevent participants from recognizing that we were interested specifically in their reactions to ignorant law violators. As they were not central to the hypotheses in this study, we did not analyze responses to the filler stimuli.

After reading the case vignettes, participants responded to the same series of questions for each stimulus (see Appendix B for a full list of questions). The dependent measures asked participants to decide whether each defendant should be convicted (a dichotomous response: yes or no) and, if they decided to convict, to indicate the severity of the sentence (on a 7-point scale: 1=the most lenient sentence for this crime, 4=the standard sentence for this crime, and 7=the maximum sentence for this crime). Although laypeople do not generally determine sentence severity, they are capable of expressing their punishment intuitions effectively and naturally through suggested sentences (e.g., Robinson & Darley 1997). Accordingly, we used sentence severity as a proxy for crime seriousness.

In order to verify that our morality manipulation was successful, we included two items that asked participants to rate the morality of each defendant's behavior (1=very immoral, 7=very moral), and to indicate how good or bad the defendant appeared to be based on the description of his or her behavior (1=a very bad person; 7=a very good person).

Results and Discussion

Unless otherwise specified, all p-values were compared to Bonferroni-corrected critical values. As most tests included three comparisons, the critical alpha value was usually 0.05/3, or 0.0133.

Manipulation Check

Participants rated the morality and goodness-badness of each defendant's behavior on two separate seven-point scales. Scores on the two scales were highly correlated, suggesting that participants treated them similarly, and that they measured a similar evaluative construct (r (15)=0.78).¹ Accordingly, we calculated a morality score for each defendant by averaging the scores on the morality and good-bad scales. As we intended, a within-subjects ANOVA showed that participants rated defendants in the moral condition (M=5.44, SD=1.16) as more moral than those in the neutral (M=4.21, SD=1.16) and immoral conditions (M=2.44, SD=0.96), F(1,16)=20.70, p<0.0001 and F(1,16)=57.61, p<0.0001, respectively. Similarly, participants rated neutral actors as more moral than immoral condition actors, F(1,16)=27.25, p<0.0001.

Sentencing Patterns

As we predicted, participants convicted immoral defendants more often (94 percent convicted) than they convicted those in the neutral (41 percent) or moral conditions (24 percent), χ² (2, N=17)=18.42, p<0.001. The conviction rates in the neutral and moral conditions were not significantly different, p>0.2.

A within-subjects ANOVA showed that immoral actors also received significantly harsher sentences (M=4.24, SD=1.52) than moral actors (M=0.47, SD=1.00), F(1,16)=85.56, p<0.001, η_p²=0.84 and neutral actors (M=0.82, SD=1.19), F(1,16)=78.92, p<0.001, η_p²=0.83. As with conviction patterns, the difference in sentence severity between the moral and neutral conditions was not significant, p>0.3. Despite the absence of statistical significance, there was a trend in the data that suggests that respondents punished neutral actors both more often and more harshly than moral actors. A follow-up multiple regression analysis in which we used participants' individual morality ratings reinforced this conclusion. This was a more powerful analysis than an ANOVA, as it took into account the possibility that some participants experienced the morality manipulations more strongly than did others. When analyzing data within the moral and neutral conditions only, we found that higher ratings of morality were associated with lower sentencing ratings, β=−0.38, p<0.05. This more powerful statistical test demonstrates that participants tended to treat moral offenders with somewhat greater leniency than they did neutral offenders.

Thus the data confirmed our expectations. Immoral actors who claimed ignorance of the law received harsher sentences than did neutral actors, who were in turn sentenced marginally more harshly than were moral actors. Although the sentencing patterns in this study matched our expectations, it is also possible that they reflected various confounds. In particular, participants may have believed that the defendants in the immoral condition should have known of the laws they broke, and that their failure to consider the legality of their actions warranted harsh sentences. In addition, participants may have considered the laws in the moral and neutral conditions to be somewhat trivial relative to the laws in the immoral condition. This may have led participants to assign more lenient sentences to the moral and neutral actors. Important to note is that both variables may have influenced sentencing independently of the morality of each defendant's behavior. We addressed this possibility in Study 2, and we enhanced the ecological validity of our conclusions by conducting Study 2 entirely between subjects.

Study 2

Method

Participants

Seventy-five Princeton University psychology undergraduates (48 females) participated in this study to satisfy a course requirement. Participants ranged in age from 17 to 30 years (M=19.86 years, SD=1.91).

Materials, Design, and Procedure

This study was similar to Study 1, but it differed in a number of significant respects. Most important, participants responded to a broader set of dependent measures (Measures 8 and 9 at Appendix B). In addition to suggesting appropriate sentences for each defendant, participants rated whether the defendant should have known of the existence of the law that he or she allegedly breached, and the importance of that law (both on a 1–7 scale). We introduced these measures in order to rule out the possibility that the sentencing patterns in Study 1 were artifacts of the particular laws chosen in each condition.

We also decided to conduct this study purely between subjects, so each participant read a series of four scenarios from the same morality condition and two filler scenarios in which the defendants were not ignorant of the laws they broke.

In the moral-ignorant condition, the defendants (a) illegally removed coral from a reef in the course of conducting research to prevent future coral bleaching, (b) violated noise pollution laws while administering a music festival for charity, (c) failed to disclose having taken a new medication before donating blood, and (d) violated fire hazard laws while attempting to house 12 homeless people. The defendants in the moral-knowledgeable condition knowingly violated the same four laws.

The defendants in the neutral-ignorant condition (a) unwittingly committed statutory rape against a 16-year-old who lied about her age, (b) illegally removed colorful bird feathers from a protected reserve, (c) removed fire-hazard notification tags from mattresses before selling them, and (d) illegally reproduced several artwork prints for personal use.

In the immoral-ignorant condition, the defendants (a) violated a customs law while importing $100,000 into the United States in the course of buying cheap heroin, (b) attempted to synthesize a “legal” alternative to methamphetamines but included one recently proscribed ingredient in the drug, (c) had sexual intercourse with a corpse, believing the law did not prohibit such conduct, and (d) advertised cigarettes on an Internet site populated by teenagers, mistakenly believing that Internet cigarette advertising was legal. (See Appendix C for the full text of all 12 descriptions.)

The between-subjects analysis was more ecologically valid than the within-subjects approach used in Study 1, as laypeople who spontaneously consider the fairness of a sentence do not have the benefit of comparing the defendant at hand to a range of other hypothetical defendants who exhibit varying degrees of moral culpability.

Finally, in addition to the three conditions in which the moral, neutral, and immoral defendants were ignorant, we included a fourth condition in which a moral actor was not ignorant. This condition was identical to the moral-ignorant condition except that the moral actor was not described as ignorant of the law. Actors in this moral-knowledgeable condition violated the same laws as the moral actors and differed only in their awareness of having violated those laws. We included this condition to determine whether it was the actor's inherent morality alone that mitigated punishment, or whether participants were indeed responding to the actor's ignorance when showing leniency. We expected that moral-knowledgeable actors would receive harsher sentences than moral-ignorant actors, thereby confirming that ignorance of the law predicts sentencing preferences.

Results and Discussion

Before conducting the following analyses, we averaged each participant's responses to the conviction and sentencing questions across the four vignettes. This step allowed us to calculate a mean sentence severity measure for each participant while ensuring that we did not artificially inflate the power of our analyses by treating participants' responses to each vignette as independent data points.

Manipulation Check

We again combined the responses to the morality and good-bad questions, as they appeared to measure the same morality construct (α=0.89). We conducted three planned Bonferroni comparisons between the three ignorance conditions, and one between the moral-ignorant and moral-knowledgeable conditions. As intended, participants rated the actors in the moral-ignorant condition as more moral (M=5.37, SD=1.22) than the neutral (M=4.03, SD=0.72), F(1,36)=20.39, p<0.01 and immoral actors (M=3.17, SD=0.94), F(1,36)=55.13, p<0.01. Similarly, participants judged the neutral actors to be more moral than the immoral actors, F(1,36)=8.49, p<0.01. As intended, participants did not rate the actors in the ignorant-moral condition as significantly more moral than the actors in the knowledgeable-moral condition (M=5.25, SD=0.65), F<1.

Sentencing Patterns

We combined each participant's conviction data to form a continuous conviction rate measure ranging from 0 to 100 percent, increasing in increments of 25 percent. We conducted analyses of variance using this measure. As in Study 1, participants in the immoral condition convicted a higher proportion of defendants (88 percent convicted) than did participants in the neutral (45 percent convicted), F(1,36)=55.14, p<0.001, η²=0.61, or moral-ignorant conditions (46 percent convicted), F(1,36)=48.75, p<0.001, η_p²=0.58. Interestingly, 69 percent of moral-knowledgeable actors were convicted, a higher rate of conviction than participants imposed on the moral-ignorant actors, F(1,36)=8.42, p<0.01, η_p²=0.19. Thus ignorance afforded moral actors some leniency, suggesting that ignorance mitigates conviction rates independently of morality.

Similarly, actors in the immoral condition received harsher sentences (M=3.29, SD=1.12) than did actors in the neutral condition (M=1.03, SD=0.71), F(1,36)=55.14, p<0.001, η_p²=0.61, or moral-ignorant condition (M=0.95, SD=0.93), F(1,36)=48.75, p<0.001, η_p²=0.58. Again, participants treated moral-ignorant actors with greater leniency than they did moral-knowledgeable actors (M=1.83, SD=1.02), imposing significantly lower sentences for the moral-ignorant actors, F(1,35)=8.42, p<0.01, η_p²=0.18.

Triviality and Obviousness of Certain Laws as Mediators of Sentencing

In this study, we addressed the possibility that the triviality and obviousness of certain laws might have augmented the differences among the morality conditions. We attempted to tackle this potential confound by asking participants to assume that all actors who claimed ignorance were indeed ignorant of the law. Nevertheless, participants may have viewed the laws in the immoral condition as more obvious, and they may have been unable to set aside their doubt about whether the immoral actors were honestly ignorant of those laws. They may also have perceived the moral and neutral laws to be trivial, resulting in more lenient sentences. Additional analyses were required to assess these possibilities.

We began by conducting a one-way, four-level, between-subjects ANOVA (moral-ignorant, moral-knowledgeable, neutral-ignorant, immoral-ignorant) to determine whether participants exposed to the three morality conditions perceived differences in the importance of the laws. We ran a similar ANOVA analyzing participants' views of the extent to which people should be aware of those laws (now referred to as obviousness), although we excluded the moral non-ignorant condition, as actors in that condition actually knew about the laws. Not unexpectedly, participants considered the immoral condition laws to be the most important and most obvious to offenders. Somewhat surprisingly, however, respondents regarded the neutral condition laws to be the least important and least obvious set of laws (see Tables 1 and 2). This pattern of responses might reflect the fact that people generally behave in a morally neutral manner, so participants identified especially well with the defendants in the neutral condition. They may have felt that laws prohibiting common acts were therefore particularly trivial and unobvious. Given the outcome of this ANOVA, we conducted two separate mediational analyses to determine the extent to which each confound contributed to participants' sentencing patterns.² We combined the morality and good-bad scales to form a single morality independent measure (α=0.86). The continuous sentence severity measure functioned as the dependent measure in both analyses.

Table 1. Mean Importance of Laws by Morality Condition

^a Denotes statistically significant difference between means, p<0.01.

^b Denotes statistically significant difference between means, p<0.10.

Table 2. Mean Expected Knowledge of Laws by Morality Condition

^a Denotes statistically significant difference between means, p<0.01.

^b Denotes statistically significant difference between means, p<0.10.

As Figures 1 and 2 demonstrate, both obviousness of the law (Sobel's Z=4.23, p<0.001) and importance of the law (Sobel's Z=2.44, p<0.01) partially, albeit weakly, mediated the effect of morality on sentencing severity. Although the mediational effects were small, we conducted a third study controlling both variables, in order to eliminate the possibility that sentencing results could be due to these extraneous factors that were independent of morality.

Note: b value in parentheses indicates the effect of morality on sentencing, controlling for the effect of assumed knowledge. ^***p<0.001

Figure 1. Mediational Effects of Assumed Knowledge of the Law on Sentence Severity

Note: b value in parentheses indicates the effect of morality on sentencing, controlling for the effect of perceived importance of the law. ^***p<0.001; ^**p<0.01

Figure 2. Mediational Effects of Importance of the Particular Law on Sentence Severity

Study 3

Method

Participants

Seventy-five Princeton University psychology undergraduates (48 females) participated in this study for partial course credit. Participants ranged in age from 18 to 22 years (M=19.46 years, SD=1.10).³

Design, Materials, and Procedure

We used the same design and procedure in this study as we did in Study 2. However, we manipulated defendants' morality in a manner that eliminated potential confounds that may have affected the results in Study 2. Specifically, in Study 2, defendants committed different crimes across morality conditions, whereas in Study 3, we fixed crime type across conditions (see Appendix D for full text of descriptions). Thus, although defendants committed the same crime in each morality condition, the extraneous circumstances that defined the morality of each defendant's behavior varied by condition. Specifically, in all cases the defendant attempted to enter the United States with a large sum of money, inadvertently breaching customs regulations. In the moral condition, the defendant planned to purchase medical equipment for sick children in Europe. Conversely, the immoral defendant intended to buy and export a large quantity of cheap heroin to Europe, whereas the neutral defendant was the manager of a foreign construction company who planned to buy inexpensive U.S. concrete. Contrary to Study 2, our results allowed us to determine whether any differences in participants' responses across the morality conditions reflected differences in the importance of each law or in the degree to which the defendant should have known of each law.

Results and Discussion

Manipulation Check

Our manipulation in this study was again successful. Responses on the goodness-badness and morality scales were highly correlated (α=0.86), so we combined them to form a global morality measure. Participants perceived the moral actors (M=6.00, SD=1.06) as more moral than the neutral actors (M=4.60, SD=0.83), t(48)=5.20, p<0.0001 and the immoral actors (M=2.60, SD=1.67), t(48)=12.56, p<0.0001, and the neutral actors as more moral than the immoral actors, t(48)=8.46, p<0.0001.

Sentencing Patterns

Contrary to the results in the first two studies, an omnibus chi-square test revealed that conviction rates did not differ significantly across the morality conditions, χ_p² (2, N=75)=5.33, p>0.05. However, a follow-up analysis in which we compared conviction rates in the moral/neutral conditions (52 percent convicted) to conviction rates in the immoral condition (76 percent convicted) showed that immoral participants were convicted more often than moral and neutral participants, χ_p² (1, N=75)=4.00, p<0.05. In addition, immoral defendants (M=2.56, SD=1.94) received harsher sentences than neutral defendants (M=0.92, SD=1.32), F(1,72)=12.23, p<0.005, η_p²=0.20 and moral defendants (M=0.88, SD=0.88), F(1,72)=15.57, p<0.005, η_p²=0.25. Thus even when we controlled crime type, participants felt that immoral defendants deserved particularly harsh sentences.⁴

Despite the consistency of these results, we conducted one final study to address a number of lingering concerns. First, we wanted to show that these results held for a nonstudent adult population, which would allow us to more strongly generalize the conclusions of our findings to the adult U.S. population at large. Second, we manipulated the defendant's ignorance, as well as his or her morality, to explore the effects of both variables on sentencing. In particular, we expected the effects of morality to be strongest when the defendant was ignorant of the law. We also controlled morality more carefully in this final study by withholding information about the defendant's background and focusing instead on his or her morality in the course of committing the specific offense.

Study 4

Method

Participants

One hundred and seventy-two Princeton University staff members (111 females) participated in this study in exchange for the chance to win one of three cash prizes. They were randomly selected from a pool of university staff and were invited to participate in this study by e-mail. Participants ranged in age from 21 to 75 years (M=41.82 years, SD=12.17). Participants varied widely in their income levels (Max=$150,000, Min=$16,000, M=$53,907, SD=23,273) and occupations (more than 50 distinct occupations represented), which suggested that the sample included people from diverse backgrounds. None of these demographic characteristics interacted with our measures, so they were excluded from further analyses.

Design, Materials, and Procedure

This study was an eight-group 2 (morality: moral, immoral) × 2 (ignorance: ignorant, knowledgeable) × 2 (crime version: customs offense, corpse removal) between-subjects experiment. Each participant saw one of the eight scenarios. We included two different crimes to ensure that the results were not specific to one particular crime (see Appendix E for the stimuli). In the customs offense scenarios, the defendant attempted to leave the United States with $12,000, either to rescue child prostitutes from Thailand (moral), or to pay for a Thai “sex tour” that included child prostitutes (immoral). In the corpse removal scenarios, the defendant removed a homeless man's corpse from the morgue either to give it a proper burial (moral) or to sell it to a biomedical research group (immoral).

In addition, we chose to eliminate the neutral morality condition from this study, as the previous studies showed that neutral and moral defendants were treated similarly. Thus defendants in each case either behaved morally or immorally. In addition, contrary to Studies 1–3, we manipulated whether defendants were ignorant of the law they were accused of breaking, in order to determine whether ignorance, in addition to morality, influenced sentence severity.

Participants who agreed to participate in the study clicked on a link in the soliciting e-mail and were taken to an Internet survey administered through Princeton University's WebSurvey platform. After giving their consent to take part in the study, participants read one of the eight scenarios. The chosen scenario was randomly selected by the WebSurvey system. Once participants read the scenario, they suggested an appropriate sentence for the defendant's behavior (0: no sentence, 1: most lenient sentence for this crime, 4: standard sentence for this crime, 7: most severe sentence for this crime). We also included manipulation checks for each of the manipulated variables, in which participants rated the defendant's morality on two scales (1: very immoral, 7: very moral, and 1: very bad person; 7: very good person) and whether they believed that the defendant was ignorant of the law he or she was accused of breaking (1: definitely not ignorant, 7: definitely ignorant). Finally, participants answered a series of demographic measures and were debriefed about the purpose of the study.

Results and Discussion

In the absence of differences in responses to the two crimes, we collapsed the data across the two crimes. Thus in the following analyses, we treated the study as a 2 (morality: moral, immoral) × 2 (ignorance: ignorant, knowledgeable) condition between-subjects experiment.

Manipulation Checks

Both the morality and ignorance manipulations were successful. We again created a morality rating score by combining responses on the goodness-badness and morality scales, which were highly correlated (α=0.84). As we expected, participants perceived the moral actors (M=5.59, SD=1.00) as more moral than the immoral actors (M=2.62, SD=1.08), t(170)=18.51, p<0.0001. Participants also perceived the defendants in the ignorant conditions (M=4.79, SD=1.85) to be more ignorant than the defendants in the knowledgeable conditions (M=3.16, SD=1.85), t(170)=5.73, p<0.0001.

We also excluded participants who failed the manipulation checks from the sentencing analyses. Specifically, we excluded participants in the ignorant and moral conditions who rated the defendants as lower than the midpoint on the ignorance and morality scales, and vice-versa for the participants in the knowledgeable and immoral conditions. This left us with a sample of 137 participants for the remaining analyses.⁵

Sentencing Patterns

As in the previous studies, we began by analyzing the conviction data, which were coded dichotomously according to whether the defendant was acquitted or convicted. A log-linear analysis on the 2 × 2 × 2 (ignorance × morality × acquitted/convicted) data matrix revealed a three-way interaction, Wald χ² (4, N=137)=19.00, p<0.001, suggesting that morality influenced sentencing differently depending on whether the defendant was ignorant of the law. To explore this interaction, we ran two follow-up chi-square analyses, one on the ignorant cells and one on the knowledgeable cells. Thus each one was a 2 × 2 (morality × conviction) chi-square analysis. As we expected, morality influenced sentences when the defendant was ignorant of the law, χ² (1, N=77)=9.29, p<0.01, Cramer's V=0.38, but not when the defendant was knowledgeable, χ² (1, N=60)=1.58, p=0.21, Cramer's V=0.22 (see Figure 3). These analyses suggest that the morality of the defendants' motives influenced participants' conviction rates only when the defendant was ignorant of the law. When the defendant was knowledgeable, conviction rates remained relatively stable and quite high.

Figure 3. Effects of Morality and Ignorance on Conviction (Study 4)

We also examined the sentences that participants imposed on the eight-point sentencing scale. Both morality and ignorance had main effects on participants' sentencing judgments (see Figure 4, which includes cell means). As we found in Studies 1–3, immoral defendants received harsher sentences (M=3.39, SD=1.89) than did moral defendants (M=1.51, SD=1.45), F(1,173)=43.88, p<0.001, η_p²=0.25. A similar pattern of results held when we conducted the same analysis excluding participants who acquitted the defendant. Even participants who sentenced the moral defendants imposed significantly more lenient sentences (M=1.98, SD=1.30) than did those who sentenced the immoral defendants (M=3.60, SD=1.72), F(1,147)=36.54, p<0.001, η_p²=0.20. This result is particularly important given that the vast majority of participants convicted the defendant in all four conditions (see Figure 3), as it suggests that even those who convicted the moral defendants showed relative leniency in sentencing.⁶

Figure 4. Effects of Morality and Ignorance on Sentencing (Study 4)

Like morality, ignorance of the law influenced sentencing decisions, such that ignorant defendants were sentenced more leniently (M=1.96, SD=1.71) than defendants who were aware that they were violating the law (M=3.25, SD=1.99), F(1,173)=18.47, p<0.001, η_p²=0.12.

Surprisingly and contrary to conviction patterns, there was no interaction between ignorance and morality, F<1. Further analysis suggested that floor effects in the moral and ignorant condition may have suppressed this interaction. Specifically, noting that the sentencing scale ranged from 0 to 7, where 4 was described as “the standard sentence for this offense,” we calculated the proportion of participants who assigned a sentence of 4 or greater in each condition. As Table 3 shows, participants used the top half of the scale in all conditions except for the moral and ignorant condition.

Table 3. Joint Influence of Motive and Knowledge of Laws on Sentencing Choices

Note: Table shows the percentage of respondents assigning sentence of 4 (standard sentence for the offense) or greater under each combination of knowledge (ignorant or knowledgeable) and motive (moral or immoral).

Accordingly, we ran an addition mediational analysis (Baron & Kenny 1986) to determine whether morality had a stronger effect on sentences in the ignorant condition. Consistent with the conviction data, we expected morality to mediate sentences more strongly in the ignorant conditions than in the knowledgeable conditions. As we expected, morality fully mediated the relationship between perceived ignorance and sentences in the ignorant conditions (see Figure 5), Sobel's Z=2.11, p<0.05, whereas morality did not mediate this relationship at all in the knowledgeable conditions (see Figure 6), Sobel's Z=0.23, p=0.82. Thus, morality influenced conviction rates and continuous sentences more strongly when defendants were ignorant of the law.

Note: b value in parentheses indicates the effect of perceived ignorance on sentencing, controlling for the effect of perceived morality. ^***p<0.001; ^*p<0.05

Figure 5. Mediation effect of morality between perceived ignorance and sentence severity when defendants were ignorant of the law

Note: b value in parentheses indicates the effect of perceived ignorance on sentencing, controlling for the effect of perceived morality. ^***p<0.001; ^*p<0.05

Figure 6. Mediational effects of morality between perceived ignorance and sentence severity when defendants were not ignorant of the law

General Discussion

As we expected, participants in the four studies dealt more harshly with immoral defendants than with neutral or moral defendants who sought to rely on ignorance of the law as an excuse for lawbreaking. In Study 1, each participant suggested sentences for moral, neutral, and immoral defendants who all claimed ignorance of the law they were charged with violating. Participants convicted immoral defendants more often and sentenced them more harshly relative to moral and neutral defendants.

However, in Study 1, participants had the opportunity to compare the morality of each defendant's behavior before assigning sentences. In reality, people tend not to have the benefit of direct comparisons between defendants of varying levels of moral culpability. Thus in the remaining studies, we achieved greater ecological validity by eliminating the artificial comparisons that may have accentuated the effects of morality on sentencing in Study 1. Study 2 replicated and confirmed our Study 1 findings that participants convicted immoral defendants more often and punished them more harshly than either neutral or moral defendants. In addition, in Study 2, we found that importance of the law and assumed knowledge of the law partially accounted for the effects of morality on sentencing.

To control for the importance of the specific law and the defendant's assumed knowledge, in Study 3 we presented respondents with scenarios in which each defendant broke the same law. We preserved the morality conditions by varying the wrongfulness of the course of conduct associated with each defendant's illegal action. Study 3 replicated the findings in Studies 1 and 2, as participants sentenced immoral defendants particularly severely and convicted them more often than moral and neutral actors. Finally, Study 4 confirmed the results in Study 3 using a more diverse sample of adult participants with two new case vignettes, and showed that both ignorance and morality play distinct roles in sentencing.

It bears repeating that these studies were not intended to replicate courtroom conditions. In common law systems, laypeople are generally not responsible for sentencing (other than the death penalty), and juries would not hear evidence on an ignorance defense unless the judge first decided to admit the defense. The studies therefore assessed conviction and punishment decisions as a means to measure laypeople's moral intuitions by asking about legal concepts that are widely familiar—conviction and punishment.

The pattern of results across the studies suggests that people particularly disfavored immoral actors, whereas they were not particularly lenient toward moral actors. It was not the case that the moral actors alone were singled out for lenient treatment; instead, the immoral actors were generally convicted more often and given a harsher sentence than were either the moral or the morally neutral actors. To the degree that these results can be generalized, this suggests that participants resisted allowing a person who acted with wrongful intent to escape punishment even if he or she was genuinely ignorant of the laws proscribing that action. If, on the other hand, a person committed an illegal act during the course of either moral or neutral behavior, then ignorance of the laws against those specific illegal acts was taken to provide a valid defense or at least a reason to mitigate the sentence given for the offense. As we expected, the primary motivation behind laypeople's intuitions about the ignorance defense is a desire to punish the blameworthy, rather than an effort to reward the exceptionally good. In other words, laypeople seemed to want to prevent the ignorance defense from being used as a shield for the cynical wrongdoer.

Generally, these results suggest that naïve sentencing patterns reflect a “subjective” approach to guilt and punishment (see Dworkin 1986; H. L. A. Hart 1958; Robinson & Darley 1998), in which there is considerable attention paid to whether the actor has either committed or formed the settled intent to commit an action that is morally wrong. According to our results, when defendants claimed to be ignorant of the law, laypeople's receptiveness to the ignorance claim varied depending on whether a defendant's illegal act was linked to a wrongful goal or motive. Similarly, we have found in other research that laypeople's sentencing recommendations for hypothetical offenders are more strongly influenced by the wrongfulness of the defendants' actions than by the harmfulness of the results (Alter et al. 2007). In other words, in an experimental setting, respondents appear to be more influenced by an actor's intent than by his or her actions or their effects.

It is important to note that there are two possible interpretations of the concept of “wrongfulness” in the context of these studies. First, a wrongful action may be construed as a legally forbidden action; second, one may consider a wrongful action as one that is morally wrong independent of legal rules. In the context of this study, people reacted strongly to the moral wrongfulness of the defendant's intentions in doing the unlawful act (consistent with Alter et al. 2007). Across the morality conditions in this study, the actors committed legally forbidden acts. However, participants were less likely to afford defendants the benefit of the ignorance defense when the act also appeared to be motivated by a morally wrongful intent, given the surrounding course of conduct. This too suggests that people intuitively adopt a subjectivist approach to sentencing. Rather than adopting a policy of absolutely allowing or disallowing ignorance as an excuse, participants' conviction and sentencing recommendations were swayed by the defendant's perceived moral wrongfulness. Thus, when the defendant acted immorally, participants paid scant attention to his or her ignorance of the law. Moral and neutral defendants, in contrast, were given substantial leeway. However, knowledge of illegality also appears to be relevant as a form of wrongfulness. We ran a condition in which moral actors knowingly broke the law, and those actors, in comparison to moral actors who were ignorant of the law, were significantly more likely to be convicted and to receive more-severe sentences.

Participants in this study appeared to sentence offenders according to the principle of just deserts, which is consistent with the subjectivist view of criminality in that it suggests that defendants should be punished according to the wrongfulness of their behavior. Researchers (e.g., Carlsmith et al. 2002; Robinson & Darley 1997) have found that when two defendants commit the same crime, differences in morally relevant circumstances matter. The more wrongful defendant, the one who embezzles to continue a life of debauchery rather than the one who embezzles to pay for medical care for a sick child, is punished more harshly. This is precisely what we found in Study 3, where all defendants committed the same act and were unaware of its illegality, but defendants received significantly harsher sentences when the act was part of an immoral course of conduct.

It is worth noting that an alternative mechanism that may explain our results is that moral defendants accrued moral credit that mitigated the perceived severity of their wrongdoing. Social psychologists have shown that people take existing moral credentials into account when interpreting subsequent, questionable behavior (e.g., Monin & Miller 2001). Similarly, moral defendants may have benefited from the so-called halo effect (Nisbett & Wilson 1977), according to which people who do good initially benefit from lingering positive associations, even when their later actions are morally questionable. Both possibilities are troubling because they suggest that criminal sentencing decisions may disregard the offender's specific illegal behavior, instead reflecting perceptions of the offender's personality. Our Study 4 shows that these mechanisms alone do not explain our results, because the claim of legal ignorance by a defendant did have some independent effect on preferred case outcomes. However, further research could explore the possibility that people employ this personality-based version of just deserts sentencing and consider whether the legal system should adopt additional safeguards to ensure that offenders are not convicted or sentenced based on conduct that is extraneous to the charged offense.

Another direction for further research is to explore whether, and in what ways, intuitions about the ignorance defense are affected by a defendant's status as a professional or an amateur. For example, it is possible that laypeople may be less receptive to claims of ignorance about home construction laws from a major real estate developer than to claims from an individual conducting repairs on his or her own home. This could arise from a view that claims of legal ignorance are less plausible from expert defendants, or a view that expert defendants have a professional duty to investigate the law and are therefore more blameworthy if they fail to do so. Further research should explore whether a defendant's perceived level of expertise affects moral intuitions about the availability of the ignorance defense.

Given that our study respondents were all U.S. residents, our analysis has similarly focused on American law. A future examination of this topic could evaluate views of ignorance of the law in other countries with civil law systems or whose common law has diverged from U.S. rules. Just as legal doctrines vary among criminal systems, it may be that moral intuitions about the ignorance of law defense also vary across countries.

A possible criticism of our results is that they are irrelevant in practical terms, because prosecutors will use their discretion to ensure that morally blameless defendants do not face charges. But prosecutorial discretion—which operates on an individual level and is not subject to appeal—should provide an emergency response to unusual cases, not a treatment for chronically unclear or conflicting legal doctrines. Moreover, the morally neutral defendants who were afforded lenience by our respondents were not nearly as memorable as the husband who steals medication for his sick wife in the well-known ethical conundrum. According to our results, the most significant division was between the morally reprehensible, who rarely were given the benefit of their ignorance, and all others, who sometimes were given the benefit. While prosecutors might exercise their discretion to avoid egregious convictions of moral paragons, there can be little assurance that morally neutral defendants handled case-by-case would receive the consistent pattern of lenience preferred by our study respondents.

Our finding that people intuitively evaluate claims of legal ignorance according to moral principles has implications for lawmakers and judges. In Liparota (1985), Ratzlaf (1994), Cheek (1991), and other cases described above, the U.S. Supreme Court evaluated claims of legal ignorance on a statute-by-statute basis. This statute-by-statute approach is not desirable because it generates unpredictability in the criminal law and invites individual judges to engage in moral line-drawing. In addition, because the Court focused on statutes in cases such as Cheek and Ratzlaf, the Court extended an ignorance defense to cynical or unsympathetic defendants based on the need to protect hypothetical future innocents. A preferable response is for appellate courts and legislatures to sift through the moral considerations that surround the ignorance of law defense and, whenever possible, to lay down principles that can be applied consistently. Our results suggest that, in contrast to the doctrine that ignorance never excuses (with exceptions for a few statutes), laypeople view knowledge of illegality to be an aspect of moral wrongfulness that is relevant to criminal case outcomes across a variety of regulatory offenses, especially when the illegal act was part of a moral or neutral course of conduct with no wrongful motivation. Essentially, our respondents seemed intuitively to treat knowledge of illegality as an aspect of mens rea, or guilty mind.

Two established doctrines are worth highlighting as examples of how legal rules about ignorance can be refined to take account of moral considerations: first, the doctrine of “willful blindness,” and second, the “official statement” exception to the ignorance maxim. Both doctrines reflect an interplay of morality and ignorance in the criminal law that is similar to our findings, and both doctrines illustrate how jurists and scholars can refine the law to reflect widespread moral intuitions. Our findings suggest that there is room to undertake such a process of refinement in American legal doctrine regarding ignorance of the law as a criminal defense.

The common law doctrine of willful blindness states that a judge can attribute constructive knowledge of a fact to a defendant, if the defendant almost knew and suspected the fact but chose not to obtain final confirmation (United States v. Jewell 1976). This doctrine reflects intuitions similar to those identified in our studies, because it reveals intolerance toward claims of factual ignorance by defendants who should have known the relevant facts. Courts have fashioned the willful blindness doctrine to ensure that criminal law can reach the defendant who is technically ignorant of a key fact but whose claim of ignorance is implausible. Put differently, the willful blindness doctrine recognizes that claimed factual ignorance may itself be an indication of moral failing. Robbins (1990) suggests that willful blindness is a form of “deliberate ignorance” that should be distinguished from true ignorance, when the defendant is morally blameless. A judge's decision to find a defendant willfully blind reflects a view that, given the circumstances of the case, the defendant's claim of factual ignorance is justifiably interpreted as a sign that the defendant either cynically relied on ignorance as a refuge or is drastically out of step with the community's norms. The doctrine of willful blindness has evolved to ensure that factual ignorance cannot be used by the cynical as a shield against conviction. It reflects a moral intuition that in some cases punishment may be deserved even when a defendant is technically ignorant of key facts.

Conversely, the official statement doctrine exists to ensure that the criminal law excuses a category of defendants whose ignorance is plainly not due to a moral flaw. The Model Penal Code's formulation of this exception states that “a belief that conduct does not legally constitute an offense is a defense to a prosecution … when: (a) the statute or other enactment defining the offense is not known to the actor and has not been published or otherwise reasonably made available prior to the conduct alleged; or (b) he acts in reasonable reliance upon an official statement of the law, afterward determined to be invalid or erroneous” obtained from one of several enumerated sources (Model Penal Code §§2.02[9], 2.04[3]). Some states have codified this exception, and courts in additional jurisdictions have acknowledged it (89 A.L.R. 4th 1026). The official statement exception is yet another reflection of a “subjectivist” view of criminality (Robinson & Darley 1998). As Parry (1997) explains, the official statement doctrine is essentially “an alternative framework for mistake of law rooted in principles of moral culpability instead of reflexive adherence to the [ignorance] maxim” (1997:5). The defense is available only when the mistake of law is reasonable, thus ensuring that it is only afforded to otherwise moral actors and is not relied upon cynically.

The official statement exception is broadly consistent with the lay intuitions revealed in our studies, which suggest that people should sometimes be afforded the right to present evidence to prove a claim of legal ignorance. This doctrine is far closer to the lay intuitions identified in our studies than the old, inflexible rule that ignorance of law never excuses. The official statement exception reflects the view that when a defendant could not have known better, the moral judgment inherent in criminal conviction is not an appropriate societal response.

At the same time, the official statement exception serves to illustrate how the criminal law could evolve farther on the basis of findings such as ours. First, the official statement exception is only available to defendants who are aware of and rely upon the law (Model Penal Code § 2.04[3][b]). In contrast, our findings show that laypeople were receptive to claims of legal ignorance when the defendant merely had no reason to suspect that a legal investigation was needed because the surrounding course of conduct was not immoral. Second, in practice, the official statement doctrine has been narrowly interpreted and applied depending on the facts of individual cases (89 A.L.R. 4th 1026). Our findings suggest that people would be willing to interpret the official statement excuse broadly when applied to defendants who were acting morally, but most narrowly when applied to immoral defendants. Finally, our findings are consistent with Segev's (2006) theoretical argument for the recognition of a defense of rational ignorance or mistake of law when a defendant relies on the statement of public officeholders or private attorneys, as well as with Segev's broader proposal that ignorance of the law should be recognized and treated as a justification rather than an excuse.⁷ In sum, although the official statement exception is significantly more nuanced than the original ignorance maxim, our findings suggest several possible directions for further elaboration that would align legal doctrine even more closely with laypeople's moral intuitions.

This article is rooted in our belief that the criminal law is more likely to command respect and to maintain its moral authority if it reflects the moral views of the surrounding community. A criminal code that is subjectivist and that punishes based on just deserts is specifically designed to conform to community moral principles. If people generally believe that an act is wrong, it will be punished accordingly. Although just deserts initially appears to be a somewhat “emotional,” and thus less utilitarian, approach to sentencing, societies that adopt the approach enjoy significant benefits. The utilitarian justification for just deserts lies in its ability to promote law-abiding behavior among its citizens (Robinson & Darley 1997). Its appeal to community intuition convinces people that their views are taken seriously, and that the legal system that dictates their behavior also advocates their well-being. Evidence suggests that people are more likely to obey the law when the legal system seems intuitively fair (e.g., Greene 2003; Nadler 2005). The doctrine of willful blindness and the official statement exception to the ignorance maxim are both examples of how legal rules can evolve to accommodate moral intuitions of this type. It is our belief that the ignorance maxim should continue to be refined in ways that reflect laypeople's sophisticated moral intuitions, and we hope that our findings about laypeople's intuitions regarding ignorance and morality will provide a sorely-needed empirical contribution to that effort.