On crime and punishment


If we were to train a machine learning algorithm to find crime, what would be the parameters that we need to feed in? What makes an action a crime?

A recent unbundling exercise urged me to find these parameters, and examine how they are evolving.

As a training neuroscientist, a Middle Eastern, a Silicon Valley resident and somone who never studied law, I found myself looking at crime with a focus on the brain, mental health, globalization and technology.

Here is what I would command my machine learning algorithm to find a crime without knowing the laws to begin with:

  • Find agents with free will
  • Look at behaviors initiated by these free agents
  • Within these behaviors, find the ones that harm others, i.e. compromise other agents’ chance of survival
  • Check if the action was intended by the free agent
  • Define it as a crime if it is illegal, i.e. against the rules of law

Regardless of how the algorithm would perform, there are emerging issues with how we find crime and how we treat it.

The basic assumption, that we all have free will, is being debated as neuroscience finds biological substances of behavior and decision making. I believe that ultimately we will remain responsible for our actions. However, two conditions, disease and disadvantage clearly disturb our ability to control our own actions. Although we came a long way toward appreciating mental disease as a disrupter of free will and intended action, we are still failing to take disadvantage into account. Childhood traumas, financial inequalities, observing and being exposed to violence, lack of a fostering family… Take an 18 year old who lives in a poor neighborhood, with a drug addict mother, and an abusive father, surrounded by violence. Is it crime in the usual sense when he fires a gunshot and kills someone? Is the 18 year old the only one who is responsible for the action?

On the other hand, is it less of a crime to harm others less directly? When a corrupt politician steals a piece of bread from a million people, or destroys the forests to create financial gains, i.e. compromises others’ survival chance while increasing his own by using his advantageous position.

Unlike free will, the illegality of crime, as defined by laws, would be assumed to be relatively stable. Even if the laws vary with morality, ethics, culture and religion, once established, they are certain and slowly evolving by nature and regulations. Crime though, is neither unyielding nor regulated. As internet technology carries the control of our belongings to the virtual space and fails to prevent anonymous actions; as globalization eases the transport of the people, businesses, and ideas, crime becomes global and intractable. Cyber crimes and global terrorist groups are obvious exemplars. We are ever more interconnected in a larger space. Yet the world’s legal systems are local and independent. The lack of a binding global legal system not only makes it difficult to detect crime, but also creates bigger gaps for unlawful organizations to thrive in weak states.


One can reasonably claim that above mentioned concerns can only be addressed with united efforts of science, technology and global powers, and not just by legal agencies. How about punishment? Are we doing a fair and constructive job in reacting to crime?

Let’s go back to our 18 year old murderer. As a consequence of his action, we are looking for a reaction that is a retribution, a deterrence for others, and possibly a kind of rehabilitation that would prevent re-offense. Meanwhile, we want the reaction to cost the least to the rest of the society.

Currently, our law focuses on retribution and deterrence. For this instance, the punishment could be several decades of captivity or life imprisonment (which would be very costly to the society) or execution. These might sound just fair, especially if we empathize with the victim. But what does it say about us, the enforcers of the punishment, and our ability to make rational and fair decisions?

Punishment, in most cases, would be a crime if it was not enforced by law. This resonates with wars, where, quite suddenly, crime becomes justified, and is committed by uniformed officials. Punishment reinforces the reinforcer, at least as much as it reinforces the offender not to reoffend.

Childhood traumas, lack of family support, and unfavorable environments affect the development of the brain and leave the developing child more vulnerable to wrongdoing. Psychiatric conditions might lead to unintended, uncontrollable harm directed to self and others. Mental health disease is just another type of disease. It should not be biased toward and not be seen as a taboo. Empathy can be taught. Rehabilitation is a powerful tool to prepare the offenders to become productive members of the society. We spend huge budgets on research to reach these conclusions, don’t we deserve to incorporate them into our legal systems instead of abiding our basic instinct for retaliation? If someone deserves to die, do we deserve to kill?

One might argue that these scientific advances suggest that the brain is deterministic and as we continue to find material correlates of behavior, the concept of free will will become irrelevant. As a consequence, there will not be any reason to treat humans any different than robots or plants. I argue against this hypothesis. Firstly, even if we were to find that human behavior is completely deterministic, we cannot rule out our capability to feel pain, experience consciousness or remember. Some people would argue that most animals do not have free will, yet this does not make us think that they are robots that we can treat them without consideration. Likewise, the value of human life is not derived from free will.

Having said that, the brain is a nonlinear system with complex functions. It is an enormous dynamic network with billions of nodes. Neither recent advances in physics nor neurobiology suggest that the functions of the brain are simply predictable. Thus it might never be possible to reliably predict future actions. With the current state of knowledge, we could most probably not predict a clinically depressed pilot to crash a plane, or a teenager to shoot someone in the head, yet alone his schoolmates. Therefore, exploring physical attributes of behavior does not seem to unveil a deterministic substrate.

Medicine has moved a long way from detecting to predicting to finally preventing diseases. If we assume that at least a subset of crimes are generated by unhealthy brain functions, we should aim to take steps toward predictive and preventive measures. It might be fair to say that the brain sciences have not progressed as speedily as other biological sciences. As a result, and also considering the complex nature of the neural systems, we might never be able to predict future actions. However we can, and we should, spend effort towards providing a nourishing environment for child development, promoting healthy social behavior, protecting mental health and preventing potential disturbances. This effort can be integrated into many services, be it education, public health, community service or urban planning.

And if we fail to do so, we should accept our responsibility of a crime committed by an unhappy, insecure, and frightened creature, and consider reducing punishment and promoting rehabilitation. As Theodore Roosevelt said “The first requisite of a good citizen … is that he shall be able and willing to pull his own weight.” Every citizen matters.

Show your support

Clapping shows how much you appreciated Aslihan Selimbeyoglu’s story.