Born bad – can biometrics decode criminality?

By Professor Fraser Sampson, former UK Biometrics & Surveillance Camera Commissioner

As we look to facial recognition and other biometrics to solve social problems, an age-old challenge keeps resurfacing.

The idea of the ‘born criminal’ has fascinated us for centuries. Since physiognomist Della Porta began using lenses to compare faces in the late 16^th century, the idea of biological determinism has dogged forensic debate.

When phrenologists offered a scientific basis for identifying criminality by skull examinations three hundred years or so later, people began studying offenders – usually retrospectively – for early warning signs in their physical makeup. As advanced biometrics like facial recognition and gait analysis are used to tackle crime, the question remains relevant today. Biometric technology is great at identifying who dunnit, but can it tell us why they dunnit?

Anatomist Franz Joseph Gall espoused a theory that our behaviour can be predicted by the topography of heads and faces, an idea built upon by Cesare Lombroso who believed he could divine criminality through reading the lumps and bumps of people convicted of crimes.

Lumping criminals together is highly dubious but the idea of being born bad continues to find support. In 2009 a US court admitted brain scans from functional magnetic resonance imaging (fMRI) into evidence in a murder trial and its creator, Kent Kiehl, still maintains that it offers a ‘promising method’ for detecting who is prone to criminality. Can he be right?

In 2019, the National Library of Medicine published a paper on ‘Biological Explanations of Criminal Behaviour’ while the Institute of Law advocates how we need to study our biological substrates to get a “fuller understanding” of criminality. It sounds impressive, however diagnosing criminality from physiology has some basic hurdles and they’re as much about the crime as the criminal. Looking for physical hallmarks of latent criminality is based on a fundamental misunderstanding of criminal law.

Offending is complex, often situational and heavily influenced by emotional, psychological and environmental elements. There are many different crime ‘types’ – crimes against people, property or the state, arising from a combination of interconnected factors and behaviours of others that were incapable of prediction. Societies produce an infinite variety of perpetrators, some committing crime for the first and only time in their lives. Few offenders confine themselves to a ‘sector’ and recidivism covers a range of offences. Are we really to believe that the faces of those who steal from shops are measurably different from those who sell drugs, traffic humans or hack computers? What about those who do all the above?

While there may be a genetic link to things like aggression which might then lead to assaults or even fatalities, a propensity to violence is not the same as having a ‘criminal brain’. Whether viewed from the outside or the inside, the skull can’t show the existence or absence of a ‘guilty mind ‘(mens rea) which is an essential element in most crimes. Some crimes need proof of intent, others require dishonesty or recklessness. Some even attract ‘absolute liability’ where there’s not much you could have done to avoid its commission whatever your dimples might have to say about it.

Genetics are proving very valuable in fields like health screening to identify risks of the person developing cancers and hereditary conditions, but these establish aetiological links between physiology and pathology, not behaviour and public policy. Laws set by governments are sociological, reflections of the mores, values and attitudes of the moment – and can change (sometimes very quickly). Things that were once crimes are decriminalised – particularly in sexual activity. Recently, the UK’s legal test for dishonesty that had stood for 35 years, changed in an afternoon, while behaviour that is a crime in some countries isn’t in others. Coupling genetics with jurisdiction is trying to link the unlinkable.

A further problem for forehead feeling and skull scanning is that research studies of ‘criminals’ are usually based on their subjects’ convictions. People displaying criminal behaviour are acquitted for non-behavioural reasons like procedural unfairness or administrative error, while others are wrongly convicted, swap charges under a plea deal or admit guilt to protect someone else. How might an MRI scan take account of fallibilities in our criminal justice systems?

Biometrics enable reliable comparison of features to confirm individual ID but using prognathous jaws and sessile ears to isolate criminal tendencies still looks like junk science. And when lawyers start pleading biological differences, we all need to be careful. If you claim to be biologically predetermined to commit crime, your behaviour may not be fixable; in some cases, the ‘illusion of scientific certainty’ might lead to a death sentence.

It would be easy to dismiss all this as biometric astrology, but it has a sinister side. Convince yourself that inherited physical features are reliable indicators of criminality and it’s a short hop to eugenics. Mapping humans is leading us towards zoemetrics – the ability to measure every manifestation of life. If the state could cut crime by screening out congenital criminals, it might claim a moral duty to do so. What use might governments make of zoemetrics to that end?

The born bad theory is an extension of the nature vs nurture debate and AI will be deployed in the continuing hunt for this biometric Snark. Biometrics attribute identity to criminals rather than identifying attributes of criminality. We might find correlations with high levels of deviant thought or low levels of internal inhibitors (both are relevant to re-offending), but we’d still be a world away from predicting conviction by physiology – and I hope we stay here. Otherwise, as Abbie Hoffman says in The Trial of the Chicago 7, people won’t be going to jail ‘because of what they did’, they’ll be going to jail ‘because of who they are’.

Emma Goldman anarchically described crime as nothing but misdirected energy. We have plenty of that in our lives and I’m not sure how much of it can be measured by our biometrics.

About the author

Professor Fraser Sampson is the former UK Biometrics and Surveillance Camera Commissioner, Professor of Governance and National Security at CENTRIC, non-executive director at Facewatch, strategic advisor to the Alan Turing Institute’s Centre for Emerging Technology and Security (CETaS), and a longtime Biometric Update columnist. He is also the author of the Springer Nature book How AI-Driven Surveillance is Changing our World: Do Robots Have the Right to Remain Silent?, which explores the legal, ethical and societal implications of AI-driven surveillance.

Article Topics

biometrics  |  criminal ID  |  facial recognition  |  Fraser Sampson  |  law enforcement