DETECTING MENS REA IN THE BRAIN.

AuthorJones, Owen D.

INTRODUCTION 2 I. DETECTING MENS REA IN THE BRAIN 7 A. Background: Initial Obstacles 7 B. The Paradigm: Eliciting Knowing and Reckless Mental States 8 C. Virtues of the Paradigm 12 D. Tools for Detecting Mens Rea 13 1. fMRI Brain Imaging 13 2. fMRI in Our Experiment 15 3. The Machine Learning Algorithm 15 4. Testing the Machine Learning Algorithm 18 E. Primary Findings 19 1. Knowing and Reckless Brain States Differ 20 2. Order of Information Matters 20 II. IMPLICATIONS OF DETECTING MENS REA IN THE BRAIN 21 A. Immediate Legal Implications 21 B. Implications for Future Work 23 III. CAUTIONS & CAVEATS 25 A. General Cautions & Caveats 26 B. Specific Cautions & Caveats 28 CONCLUSION 30 INTRODUCTION

Mental states matter. Consequently, we and colleagues designed and executed a brain-imaging experiment attempting to detect--for the first time--differences between mental states relevant to criminal law.

Imagine you've just killed someone in Colorado. It was not your purpose or desire to kill him. Nevertheless, another human being is dead. Arrested and on trial, you do not dispute that your action unjustifiably caused his death. But whereas the prosecutor argues that you knew someone would die as an inevitable by-product of your actions, you assert in your defense that you knew no such thing. Instead (you claim) you were merely reckless. That is, you acted as you did with awareness of a substantial risk that someone would be fatally injured, but without knowing you would kill anyone.

In Colorado, as in many states, there is a huge difference in the sentencing ranges for those convicted of knowing and reckless homicides. In Colorado it means the difference between being sentenced to sixteen to forty-eight years in prison and none. (1) So your fate rests in the hands of lay jurors who will decide what your mental state was at the time of the fatal act. Specifically: Did you know you would kill someone, or were you merely aware of a risk that you would?

Now, any plausible theory of the point or purpose of meting out punishment to offenders--whether utilitarian, retributivist or expressivist--will recognize good reasons to condition punishment, or its amount, on the offender's mental state. Mental states matter to the nature and severity of incentives to which human behavior is sensitive, to moral desert, and to society's collective outrage. But, whatever its rationale, the practice of predicating differences in punishment on differences in mental state means that you now face two large problems ignored by our current criminal justice system.

First, the criminal justice system simply assumes that most jurors can reliably distinguish between the two mental states at issue. It is well known to all first-year law students that the supermajority of states follows the Model Penal Code's long-standing approach to categorizing culpable mental states into four types: purposeful, knowing, reckless, and negligent. Large numbers of offenses, including homicides, are then subdivided into corresponding categories. And the extent of legal intervention--length of prison stay, for instance--scales accordingly. But less well known is that a body of experimental evidence suggests that jurors are not particularly good at understanding which category is which. (2) Subjects frequently get it wrong, even when directly instructed on the relevant legal definitions and standards. They find it particularly difficult to sort defendants between the Model Penal Code's categories of "knowing" and "reckless." They confuse the two about 50% of the time, under some conditions, and do little better under others. (3) In such a case, that's nearly coin-flipping odds of false conviction.

The other big problem is that no one knows if the legally assumed and statutorily instantiated distinction between knowing and reckless mental states reflects an actual and inherent psychological difference. Sure, we might all believe such a difference exists, on the basis of introspection alone. But introspection seems an insufficiently sound basis on which to lay the very foundations for policy distinctions of such large consequence. The supposed distinction between knowing and reckless could be nothing more than a convenient fiction, upon which countless trials--and far more plea bargains--have been built.

To see the bite of this second problem, first consider the way in which Legal Realists have approached the mens rea categories. The Legal Realist tradition is well-known for the claim that many legal terms and concepts falsely purport to classify defendants and their circumstances on the basis of their intrinsic features. (4) Instead, assert the Realists, defendants are classified on the basis only of the judge's desire to hold some liable and to decline to hold others liable, even when those two groups of people do not differ in any way other than in the eye of the judge. (5) But then note that this critique has extended to juries as well. That is, the Legal Realists claim that the question of whether you were knowing or reckless when you performed the act that killed someone only purports to be a question about your psychology.

The important, long-lingering question--as relevant to retributivists as to utilitarians--is therefore this. Does the distinction between Model Penal Code mens rea categories, such as knowing and reckless, reflect an intrinsic psychological difference, actually found in human beings? If so, we believe that one should expect in principle that there would also be a difference between the brains of reckless and knowing individuals, at the times of their actions. Because, after all (and setting aside some philosophical subtleties (6)) anytime there is a psychological difference there must also be a brain difference.

So is there a neural difference or not? Regardless of where legislators choose to draw the lines between different mental states, and regardless of how many categories they create, those categories should reflect something real, and not merely assumed or hoped-for. Because the lives and liberties for thousands, each year, depend on the mental state category assigned to them.

So if we have used the supposed distinction between knowing and reckless (and other Model Penal Code categories) to justify a different treatment under the law, when there is in fact no detectable or meaningful psychological distinction, then widespread injustice will have followed in the wake of the Model Penal Code, and will continue indefinitely, if unchecked.

With a grant of nearly $600,000 from the MacArthur Foundation Research Network on Law and Neuroscience (7) (Research Network), we--as part of a larger interdisciplinary team (8)--set out to investigate the knowing-reckless distinction in the brain, and the boundary that may separate them. Specifically, we aimed to see if we could use brain activity alone to detect the difference between those who the law would classify as "knowing" and as "reckless." By combining the relatively new technical achievements of functional magnetic resonance imaging (fMRI) with new advances in the analytic abilities of machine-learning algorithms (a form of artificial intelligence) our team conducted the first assault on this thorny legal problem.

This Essay reports and describes, for a legal audience, the results and implications of our experiment. We found evidence strongly supporting the existence of a brain-based distinction between knowing and reckless mental states. Our detailed neuroscience paper was first published in a dedicated peer-reviewed science journal, the Proceedings of the National Academy of Sciences. (9) It received some sensationalist press coverage, including headlines such as this one from the British Daily Mail: Something On Your Mind? AI Can Read Your Thoughts and Tell Whether You are Guilty of Committing a Crime. (10) Half-truths like these are dangerous. So our goal here is to explain for a legal, non-scientific audience what we did and--more importantly--how it does and does not matter for the law.

Our team's discovery is relevant to law in two ways. (11) First, it provides new information relevant to the substantive debates over the accuracy and legitimacy of the distinction the Model Penal Code draws between knowing and reckless mental states. In this one important domain, that is, there is reason to think that what the law purports to do--draw a distinction in liability on the basis of a distinction in psychological state--is what it actually does. Second and collaterally, our results serve as a concrete and salient example of how new neuroscientific techniques, sometimes partnered with artificial intelligence tools, can be used to probe matters of legal relevance.

We proceed in three primary parts. Part I provides an overview of the experiment and the results. Along the way, it offers a necessary but brief and accessible introduction to how fMRI brain imaging works. Part II discusses the important implications of this new finding. Part III provides necessary caveats and cautions, to ensure that our results won't be over- or misinterpreted.

  1. DETECTING MENS REA IN THE BRAIN

    1. Background: Initial Obstacles

      The central challenge was to develop an experimental paradigm that could elicit states of mind that legal scholars and the legal system in general would consistently classify as knowing and reckless. So suppose two different scenarios in which a person takes something that does not belong to him without permission. In both scenarios, our defendant emails his neighbor to ask permission to borrow that neighbor's car. Assume further that, in the past, the neighbor has said yes to this request about half the time, and about half the time has said no.

      In the first scenario, our defendant checks his email and sees that the neighbor has said no. Then the defendant borrows the neighbor's car anyway, knowing full well that he does not have permission to do so. He figures (incorrectly as it...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT