Offending Nature?
Might voluntary DNA analysis after a conviction one day be useful and workable in some societies?
Researchers in the United States recently produced a paper for publication entitled; “A Deep Neural Network Model to Predict Criminality Using Image Processing”. The paper seeks to use facial recognition technology to identify a person’s predisposition to crime. The paper includes:
“Identifying the criminality of [a] person from their facial image will enable a significant advantage for law-enforcement agencies and other intelligence agencies to prevent crime from occurring.”
The publisher, Springer, withdrew the paper following a campaign by scholars concerned about its racial assumptions, who wrote: .
“Machine learning programs are not neutral; research agendas and the data sets they work with often inherit dominant cultural beliefs about the world. These research agendas reflect the incentives and perspectives of those in the privileged position of developing machine learning models, and the data on which they rely. The uncritical acceptance of default assumptions inevitably leads to discriminatory design in algorithmic systems, reproducing ideas which normalize social hierarchies and legitimize violence against marginalized groups. Such research does not require intentional malice or racial prejudice on the part of the researcher. Rather, it is the expected by-product of any field which evaluates the quality of their research almost exclusively on the basis of “predictive performance”.
The authors of the paper, likely because it is not their forte, did not show any interest in how their proposal would have profound effects on societies. And yet by arguing for application of their research to public policy they went well beyond an exploration of their research domain and entered into advocacy on matters of social policy and politics.
The United States in particular is a mass incarceration society home to a quarter of the world’s prison population; well over two million people. The decades following the 1960s civil rights legislation saw a 500% increase in the number of prisoners.
Today, the United Kingdom has a few dozen prisoners subject to a lifetime tariff and who will therefore never be released. This is the highest in Europe, where many countries allow no such status; the European Court of Human Rights views a lifetime tariff without the possibility of review as inhuman and degrading. The United States, by contrast, has over 50,000 prisoners who will never be released and states like Texas rationalise the practice as providing the claimed social benefits of execution without the actual execution.
France has a few hundred prisoners serving life with the possibility of parole; other than the United Kingdom which at over 8000 has more proportionate to the prison population than even the United States, this is the norm for Europe. The United States has over 200,000.
In addition to the around seven million people in the US justice system either in prison or on some form of supervision, millions more are awaiting trial.
Crucially, every single category above is dominated by people who do not identify as white. In some parts of the United States, the majority of Black men under 30 are presently somewhere in the justice system. More broadly, around a third of all US citizens who identify as Black have a felony conviction, with almost half of those having spent time in prison.
Add to all of this the present contentiousness surrounding US policing and much wider concern amongst people of all political stripes about surveillance societies — London is by far the most surveilled major international city in the world — and the excitement at Harrisburg University at an explosive social proposal seems profoundly ill-advised at best.
All of this sets into the political and social context the possible application of genetics to predicting future criminal behaviour. At face value, it seems like a no-hoper. And yet, with appropriate safeguards — in itself a major issue — might there nevertheless be a possible place for the well-targeted application of good science to the huge social problem of re-offending?
A comfortable majority of people released from prison in the UK re-offend. In the US, over three quarters of all released prisoners re-offend within five years. The policy environment stresses environmental causes and solutions alone. The justice system does, however, recognise causes which extend from nature rather than nurture. Neurocriminology, for example is a fast developing field of research.
It is sometimes assumed that identifying natural causes of offending leads to mitigation pleas in court and a denial of personal responsibility; this in turn lends a pinkish political hue to the notion of natural causes. Yet the opposite judicial trend equally likely to be true. A judge told that an offender is naturally predisposed to serious re-offending may decide on a longer sentence for greater public protection since re-education or a change in environment may not have much effect.
The wholesale application of DNA analysis to citizens with no criminal record is, for reasons related to the problems related to the surveillance society details given above, likely a political non-starter. But what if actual offenders were offered the opportunity to have their DNA analysed in order to help inform tailored programmes aimed at reducing thier likelihood of re-offending?
At present, offenders are typically profiled using qualitative interviews and relatively crude proformas. Most systems operate primarily on the basis of actuarial principles rather than individual analysis. This is partly related to the complex interwoven causes of crime, but also to resource scarcity. Proposals for new forms of anti-recidivism intervention must be highly cost-efficient and not be based upon assumptions of long term savings from lower crime rates (treasury officials are, for solid practical reasons, usually unmoved by such future savings arguments).
Genetic analysis of offenders would be cheap; the main cost would likely come from translating the results into workable personalised development programmes. The societal context would be critical, of course; it is hard to see any appliction in the present United States for example. But might there be room for an early pilot elsewhere under the auspices of a research team which contained both scientists and social scientists? Perhaps it wouldn't be scalable in the end, but it might. A worthwhile potential avenue of research?
