A scan says nothing about this one person

A printout lies on the table: a section through a head, one region picked out in colour. A tumour, an old injury, a conspicuous activation. The image seems to weigh more than the rest of the file — more than the witness who remembers, more than the diagnosis that came out of conversations. It looks like the layer beneath all the other layers.

But depth is not a legal category. Every action, intentional and unintentional alike, is made possible by a brain. If neural causation on its own ruled out responsibility, nobody would be responsible. The image on the table would then prove no more than that the defendant is a human being.

What the law actually asks

The wording is narrower than the debate usually assumes. Section 20 of the German Criminal Code acquits nobody because a cause has been found:

Every word in it is working. The reason must come from a narrowly drawn list: pathological mental disorder, profound disturbance of consciousness, intellectual disability, or severe other mental disorder. Section 21 governs the intermediate case: if either of these two capacities is substantially diminished, the sentence may be reduced.

Two tests are contained in this, not one. First a condition from the listed catalogue must be present. Then — and this is the real point of contention in the courtroom — that condition must have damaged one of two capacities: the appreciation of wrongfulness, or the control of one’s own conduct in accordance with that appreciation. Both refer to a moment in the past.

A brain image can contribute substantially to the first test. A tumour, epileptic activity, a dementia or a documented injury are findings, not interpretations. To the second test the same image contributes at most a piece of circumstantial evidence. It shows a head at the moment of the recording, not a mind at the moment of the offence.

Explanation is not excuse

A finding may explain why impulse control was impaired without settling how far that excuses. Conversely, an unremarkable scan proves no full capacity. Many mental functions arise in networks and situations; single images indicate differences between groups better than the mental state of one person on a past evening.

The legal scholar Stephen J. Morse has a name for the overvaluing of neural evidence: “brain overclaim syndrome”. He means the reflex of turning a mechanism straight into a legal consequence. Particularly risky here is the inference from today’s image back to the time of the offence. Medication, sleep, substances, development and context may all have changed. Correlations with aggression or reoffending are not mind readers.

Michael S. Pardo and Dennis Patterson take the objection one level deeper. Their charge is not empirical but conceptual: sentences such as “the brain decided” confuse the part with the whole. Decisions, intentions and knowledge are predicates of persons, not of brain regions. Whoever shifts them has not found a new result but changed a category.

From group finding to individual case

How far a statistic carries can be played through on one of the field’s most cited papers. Eyal Aharoni and colleagues studied 96 released offenders who lay in a scanner during a task on impulse control. Those who showed little activity in the anterior cingulate cortex when making errors had roughly twice the odds of being rearrested within four years — holding other observed risk factors constant.

A year later the same authors examined how well this transfers to individual people. They tested the discrimination and calibration of their predictions against conventional models and found “moderate to strong” accuracy. That is more than critics had expected, and still something other than a statement about a person.

For doubled odds are a statement about distributions. The groups continue to overlap heavily; most people with the unfavourable feature are not rearrested, and some of those without it are. An effect size of this magnitude is remarkable for social science and too coarse for an individual forecast. Above all, though, it answers the wrong question: the court does not want to know what this person will do in four years, but what they were capable of on a particular evening.

What the colourful image does to lay people

A second argument against brain images in trials holds that they dazzle jurors and lay judges. In 2008 David P. McCabe and Alan D. Castel reported that an accompanying brain scan made scientific explanations appear more credible. The finding became the standard argument of the cautionary literature.

It has only partly survived scrutiny. Robert B. Michael and colleagues set out to build on the effect, failed to do so, and instead carried out ten systematic replications with almost 2,000 people. In a meta-analysis of old and new data, little or no influence of the brain image remained. Independently of this, N. J. Schweitzer and colleagues found with simulated jurors that images changed nothing in a defence against a charge of intent; the subtitle of their paper is their result: “No impact”.

That shifts the worry rather than dissolving it. The danger evidently lies less in the image than in the sentence accompanying it — in the claim that something has been measured here that answers the court’s question.

Where the evidence actually lands

In practice, neuroscientific material is submitted in support of acquittal less often than the debate suggests. Nita A. Farahany analysed US judicial opinions from 2005 to 2012 in which defendants had introduced neuroscientific or behavioural-genetic evidence. It served increasingly to challenge competence to stand trial, to attack the performance of counsel at trial, and to mitigate the sentence. The centre of gravity therefore does not lie with the question of guilt, but behind it.

Brain research had its most visible effect where it spoke about a whole group instead of about one defendant. In three decisions within eight years, the Supreme Court of the United States relied on research into the maturation of the adolescent brain. In 2012 it barred the states from mandating life imprisonment without the possibility of parole for offences committed by minors. Laurence Steinberg, who has examined this case law, draws a sobering conclusion from it. The neuroscientific evidence probably persuaded not because it revealed anything new about adolescence, but because it agreed with common sense and with behavioural research.

What remains

Two things hold. First, the law asks about capacities in a moment, not about causes as such — and no measuring device can answer that question, because it contains an evaluation. Second, a brain finding is not thereby worthless: it belongs alongside records, behavioural history, clinical interview and informant reports, and a good assessment names alternative explanations and uncertainty instead of turning an image into a verdict.

What is contested is how long this division of labour will hold. In 2004 Joshua Greene and Jonathan Cohen predicted that neuroscience would change the law — but not by refuting its doctrine. It would shift people’s moral intuitions, away from retribution and toward a consequence-oriented punishment. Morse, and Pardo and Patterson, hold against this that the conceptual foundations remain untouched by that. The dispute is open, and it is not an empirical one.

In everyday life the fallacy shows itself in a missing specification. Where someone says the brain caused the act without saying which capacity is supposed to have been impaired, at what time and to what degree, an explanation has been dressed up as an excuse. The scan is then no acquittal in colour. It is an image that nobody has yet translated into a question.