The public defender was worried. “If I show you my client’s confession, would you be able to tell me if what he is saying is correct?”
This is not an unusual request. Suspects frequently make damaging statements, whether they committed a crime or not. The police rely on videotaped interrogation and Miranda warnings to get confessions admitted as evidence at trial because of the simple, widely held, false belief that people don’t confess to crimes against their own interest.
But is that belief supportable by facts? Innocence Project data show that over a quarter of those exonerated through DNA evidence — i.e., people who were convicted of crimes that later scientific evidence proved they did not commit — had under police interrogation confessed to the crime. Some interrogation tactics (the Reid technique, for example) have been criticized for their propensity to elicit false confessions. Mentally ill and non-neurotypical suspects with low IQ seem to be particularly vulnerable.
When I’m investigating a violent death and the police have elicited a confession from someone they have arrested for that crime, the cops who attend the autopsy might tell me about the confession — but, either way, I don’t routinely watch video from interrogation rooms.
“You have the confession on video?” I asked the public defender. She nodded and reached over to my computer to pop in the CD.
I — like other forensic pathologists — am no expert about false confessions. Nevertheless, the first thing I noticed was that the suspect was asleep in the chair. The officer performing the interrogation had to keep waking him up. There was a point where I saw drool coming from the guy’s mouth. He wiped at it with the back of his hand, listlessly.
He wasn’t answering the questions in any detail — just saying “yeah” or “yes” over and over again to whatever he was asked, and then dozing off some more. It wasn’t clear to me that he was even hearing the questions. It looked like he wanted to answer them affirmatively so they’d stop coming and he could sleep.
“What’s wrong with him?” I asked the public defender. She shrugged. “Is he a drug user?”
“Yes,” she said.
“Big time. Alcohol too. He was just coming off a 3-day bender.”
“He’s in withdrawal. That’s why he’s drooling and dozing off. He has no idea what they are asking him.”
The lawyer was floored. She had come to see me so I could tell her whether her client’s confession matched the physical findings on the body, but as far as I could see, her client had provided no information at all. He just mumbled affirmations of whatever the police asked him.
The police officers were supplying details drawn from what they knew from the scene investigation and the autopsy. They supplied the facts they wanted, strung them together, and got their prime suspect to say yes, he had made those things happen. It wasn’t a confession. It was a textbook demonstration of investigators contaminating a witness with what they already knew. On video.
I referred the public defender to a forensic toxicologist, and he confirmed what I had suspected: All the defendant’s observed symptoms were classic for withdrawal. The public defender requested a court hearing to get the confession thrown out, and the forensic toxicologist testified. The district attorney dropped the charges, and I never got called to court. But I will never forget that case.
Awhile later I had a far different case, a complex one, with a burned body. When I performed the autopsy, it was unclear based on the anatomic findings whether the death was due to strangulation, thermal burns, or a combination of both.
The police later told me they had a suspect in custody, and that he had confessed to strangling the decedent. This time I asked to see the video. The suspect was awake, alert, and cooperative, but the audio was hopelessly corrupted. I couldn’t hear a thing he said, even after the sound was digitally enhanced. I refused to consider the suspect’s statement in my opinion at all. The DA went ahead with the case, and I later testified in court to my equivocal autopsy findings. The suspect got convicted.
So should medical examiners be told there is a confession in a case they’re investigating? If yes, when? After they have done the autopsy, or before? Or, not at all? Should officers who know the autopsy or scene findings be interrogating persons of interest or witnesses at all?
Should medical examiners rely on confessions to inform their death certificates or testimony? What if the confessions are coerced? What if they’re false?
It’s well past time we gathered some answers to these questions.
I was one of the seven co-authors of a recent attempt to start, with a study published in February in the Journal of Forensic Sciences. It included a survey targeted at members of a professional organization of medical examiners, coroners, and forensic pathologists.
We presented a hypothetical case of traumatic infant head injury, and changed two potentially biasing pieces of contextual information without altering any of the medical autopsy findings. The two pieces of information were the race of the child (Black or white) and who the caregiver was who found the injured child (mother’s boyfriend or child’s grandmother).
The study showed that while the majority of respondents chose the manner of death as underdetermined, 41% were confident enough to decide whether the manner of death was either accident or homicide.
Those who were given the scenario of a Black child found by the mother’s boyfriend were more likely to choose homicide as the manner of death. In contrast, those given the identical medical data with a white child found by the grandmother were more likely to determine that the death was an accident.
Why is that important? Because the determination of manner of death — whether an injury is the result of an accident, a suicide, or a homicide — has serious legal implications. The survey scenario did not include a confession, but I wonder what the results would have been if it did. I hope I get to study that someday.
We need to continue to study bias in forensic pathology. Our entire discipline is based on the assumption that by going to the scene, by seeing everything and being exposed to the police investigative findings first, we can make an accurate decision about the autopsy findings.
But what if aspects of the police investigation actually lead us to the wrong conclusions? What if evidence was missed? What if the eyewitness made a mistake? What if the confession was coerced or influenced?
Forensic scientists need to know whether extraneous circumstantial information (like race, caregiver, the presence of a confession), or whether working closely with the police can improperly influence our diagnoses and determinations.
In the past, forensic pathologists have shown a great deal of resistance to the study of bias in our field. I hope this newly published study changes that. I hope it will not be the last one to ask questions about what information is crucial for the forensic pathologist to have, and what information is more likely to introduce bias.
I also hope that, in time, forensic pathologists will accept (like DNA scientists and fingerprint examiners before them) that they are not immune to bias, and that we need to develop better practices and protocols to minimize it.
Ultimately we are seeking the truth — in a very high-stakes environment, where justice and grave injustice are in the balance.
Judy Melinek, MD, is a forensic pathologist and CEO of PathologyExpert Inc. She is currently working as a contract pathologist in Wellington, New Zealand. Her New York Times bestselling memoir, co-authored with her husband, writer T.J. Mitchell, is Working Stiff: Two Years, 262 Bodies, and the Making of a Medical Examiner. The duo have also embarked on a medical-examiner detective novel series with First Cut, available from Hanover Square Press.