Sorting Guilty Minds

Francis Shen & Morris Hoffman & Owen Jones & Joshua Greene & René Marois

Posted in ,

Every crime consists of an illegal act committed with a guilty mind. Proving the act alone is rarely enough because the law typically does not criminalize accidents. Yet while the distinction between accidents and non-accidents seems straightforward, punishing the “guilty mind”—as students quickly learn and practicing attorneys well know—is much more difficult.

For the last fifty years, the Model Penal Code (MPC) has attempted to ease this difficulty by organizing guilty minds (or “mens rea”) into four categories: purposeful, knowing, reckless, and negligent. Each of these four kinds of culpability is meant to describe distinct and legally relevant relationships between the wrongdoer and the harm he causes.

Specifically, a purposeful wrongdoer causes harm when he engages in conduct with the desire to cause the harm. A knowing wrongdoer does not desire to cause the harm, but knows it is “practically certain” that the harm will occur in the course of him achieving some other desired result. A reckless wrongdoer does not desire to cause the harm, but is consciously aware he is taking a substantial risk of causing that harm. A negligent wrongdoer also takes a substantial risk of undesired harm, but is not consciously aware of that risk. When the MPC—or other state criminal code—imposes different penalties for commission of the same act with the same harm but different states of mind, it punishes purposeful crimes more severely than knowing ones, knowing ones more severely than reckless ones, and reckless ones more severely than negligent ones.

This four-part hierarchy of guilty minds was a great advance in the criminal law when it was introduced, and it remains highly influential in a majority of the fifty states. Yet for all their assumed value the MPC’s provisions on mental states have almost entirely escaped the scrutiny of empirical evaluation. Fifty years into the MPC’s reign, we still don’t have a good answer to the basic question: Can jurors actually see beyond the basic distinction of accidents and non-accidents to make the finer-grained distinctions between purposeful, knowing, reckless, and negligent behavior that the MPC requires of them? The Article on which this essay is based, Sorting Guilty Minds, puts the MPC’s assumptions to the test.

With a grant from the John D. and Catherine T. MacArthur Foundation’s Law and Neuroscience Project, we conducted a series of experiments that would answer two fundamental questions: 1) Do ordinary people punish crimes in the way that the MPC assumes?; and 2) if not, is it because they cannot distinguish among the four MPC categories to begin with? Within the first question we also tested whether having the MPC culpability definitions available to subjects (akin to the instructional definitions a jury would be given) improved their ability to distinguish punishment between the MPC categories.

We asked subjects to read short written scenarios depicting the commission of a crime by a fictional protagonist named “John.” We varied fact patterns by creating thirty different themes, and within each theme we created five scenarios—varying John’s mental state across the four MPC mental state categories and a fifth category of blameless. To provide some external validation for our constructed scenarios, we asked nine law professors to read the scenarios and identify the mental state of the protagonist. These law professors were able to identify the correct mental state with nearly 90% accuracy. But what about our lay subjects?

The results of our experiments, described in detail in the Article, are both gratifying and disturbing. With one important exception, subjects were very good at punishing according to the severity hierarchy of the MPC. As can be seen in Figure 1, across all of our themes (which we designed to cover the spectrum of low, medium, and high harm) subjects consistently and robustly punished purposeful acts more severely than knowing acts, reckless acts more than negligent acts, and negligent acts more than blameless acts. As harm went up, so did punishment, but the separation between purposeful/knowing, reckless/negligent, and negligent/blameless remained. This separation is remarkable, especially considering the fact that these subjects were given no instructions and were asked simply to assign punishment levels.

Figure 1. Average Punishment Ratings For Purposeful, Negligent, and Blameless Scenarios from Experiment 1 (“How Do Subjects Punish With No MPC Instructions?”) (Plotted By Harm Level Ranking of Theme)

(click to enlarge)

What To Notice in Figure 1: The average punishment ratings for purposeful, negligent, and blameless scenarios are completely distinct from one another. That is, they do not cross, and their order is consistent with the assumptions of the Model Penal Code.

These results suggest that two of the MPC’s four culpability categories—purposeful and negligent—resonate deeply with our intuitions of blame and punishment. Of course, findings in psychology have already demonstrated that we generally blame intentional wrongs far more severely than accidental ones. But our results show that the MPC accurately reflects these general moral categories of “intention” and “accident” in its specific definitions of purposeful and negligent.

The bad news for the MPC is that our subjects were much worse at distinguishing between knowing and reckless conduct. We found that subjects could correctly identify knowing conduct only 50% of the time, and reckless conduct only 40% of the time. Their punishment decisions mirrored this confusion. Figure 2 shows the punishment ratings from Figure 1 overlain with the average punishment ratings for knowing and reckless. Figure 2 clearly shows that there is no systematic difference between the two rating lines, with average knowing and reckless ratings often crisscrossing each other. Statistical analysis confirms that subjects routinely failed to punish knowing conduct more harshly than reckless conduct. Subjects were no better at distinguishing their punishment between these two categories when they were given the MPC definitions either at the beginning of the experiment or continuously throughout the experiment.

Figure 2. Average Punishment Ratings For Knowing and Reckless Scenarios for Experiment 1 (“How Do Subjects Punish With No MPC Instructions?”) (Plotted By Harm Level Ranking of Theme)

(click to enlarge)

What To Notice in Figure 2: The average punishment ratings for knowing and reckless scenarios cross each other repeatedly. The figure visually presents what is confirmed by statistical analysis discussed in the Article’s Appendix A: There is no significant difference between punishment ratings of knowing and reckless scenarios. That is, punishment ratings of knowing and reckless actors are not consistent with the assumptions of the Model Penal Code.

Figure 2 also suggests that if we were to collapse knowing and reckless into a single category, subjects would be able to consistently identify this single category clearly as distinct from purposeful and negligent, and to punish it accordingly.

Results from additional experiments—in which we explicitly asked subjects to identify scenarios as describing a purposeful, knowing, reckless, negligent or blameless act—help to explain the punishment findings. We found that subjects could correctly identify 78% of purposeful scenarios, 88% of blameless scenarios, and 48% of negligent scenarios. When knowing and reckless scenarios were presented, subjects identified them as either knowing or reckless 80% of the time. On the one hand, it is quite impressive that ordinary people are just about as good at identifying purposeful behavior, and identifying knowing and reckless behavior as one or the other, as they are at identifying blameless scenarios. On the other hand, subjects’ significantly worse performance at distinguishing knowing behavior from reckless behavior raises important questions about the utility of these two categories, and the propriety of continuing to deploy them.

Perhaps subjects’ failure to distinguish knowing from reckless behavior may not be all that problematic. After all, there are very few crimes that depend on a jury’s ability to distinguish these two states of mind. Perhaps this reflects legislators’ own sense that there is no moral difference between these two kinds of culpability. On the other hand, in many states there is one very important kind of crime—homicide—for which the difference between knowing and reckless can have enormous consequences. In Colorado, for example, a knowing homicide is called second-degree murder and is punishable by a mandatory prison sentence of between sixteen and forty-eight years. But if the same killing had been done recklessly, it would be called manslaughter and would carry a non-mandatory sentence of between two and six years.1 There are many states with similar provisions.

This is not just a theoretical difference. It is not at all uncommon for jurors in homicide cases to be instructed that there are varying levels of homicide and that they should choose one to apply based on the defendant’s state of mind. If, as our experiments suggest, people have difficulty distinguishing between knowing and reckless behavior generally, then it is likely that jurors have the same difficulty distinguishing between knowing and reckless homicide.

The results of these findings are thus potentially profound. If the knowing/reckless findings reported here are confirmed in subsequent studies, and if we as a society value treating similarly situated defendants alike (or at least non-arbitrarily), then we need to do a better job of defining these two categories.

There are three types of responses. First, we could rewrite our jury instructions to emphasize that knowing conduct is not so much about risk-taking as it is about the harmful side effects that a wrongdoer is willing to cause in order to achieve some other desired purpose. But such a change might have the unintended consequence of conflating purposeful and knowing states of mind. Second, if we keep knowing as a risk-based state of mind, perhaps we should change our definition of reckless to emphasize that it is a lower risk than knowing’s “practically certain” standard. Third, and more aggressively, we could simply abandon the distinction between knowing and reckless conduct, at least in the few kinds of cases—such as homicide—where this seemingly impossible distinction drives dramatic and potentially arbitrary differences in punishment.

Whether reforms such as these will work requires more inquiry. But our results suggest that such further inquiry is sorely needed. Although the MPC does an admirable job in many ways, if jurors really cannot distinguish knowing and reckless behavior, then significant punishment differences based on that distinction are profoundly unjust.


Copyright © 2011 New York University Law Review.

Francis X. Shen is a Visiting Assistant Professor at Tulane University Law School & The Murphy Institute. Morris B. Hoffman is a District Judge for Second Judicial District (Denver), State of Colorado; Adjunct Professor of Law at the University of Colorado; Research Fellow at the Gruter Institute for Law and Behavioral Research; and Member of the MacArthur Foundation Research Network on Law and Neuroscience. Owen D. Jones is the New York Alumni Chancellor’s Chair in Law & Professor of Biology at Vanderbilt University; Director of the MacArthur Foundation Research Network on Law and Neuroscience. Joshua D. Greene is an Associate Professor of Psychology and Director of the Moral Cognition Laboratory at Harvard University. René Marois is an Associate Professor of Psychology and Director of the Human Information Processing Laboratory at Vanderbilt University.

This Legal Workshop Editorial is based on the following Law Review Article: Francis X. Shen, Morris B. Hoffman, Owen D. Jones, Joshua D. Greene, & René Marois, Sorting Guilty Minds, 86 N.Y.U. L. REV. 1306 (2011).

  1. COLO. REV. STAT. ANN. §§ 18-1.3-401(1)(a)(V)(A), 18-1.3-406(1)(A), 18-1.3-406(2)(a)(V)(A), 18-3-103(1), 18-3-103(3)(a), 18-3-104.

Post a Comment (all fields are required)

You must be logged in to post a comment.