Self-incrimination and confessions

Desjardins plays the class two short clips from training interviews, with the suspects' faces blurred. In the first, an officer asks open questions, listens, and lets long silences sit until the suspect fills them. In the second, an officer leans in, tells the suspect the evidence is overwhelming, says the judge will go easy on someone who shows remorse, and keeps going for hours. Both interviews end with a confession. He asks the class which confession they would trust. Every hand goes to the first. Then he tells them the second suspect was later proven innocent by DNA. The law of confessions, he says, exists because the second kind of interview produces confessions from innocent people far more often than anyone expected before DNA testing showed it.

The privilege against self-incrimination

The Fifth Amendment says no person shall be compelled in any criminal case to be a witness against himself. The privilege protects testimonial evidence, meaning communications that reveal the contents of a person's mind, not physical evidence such as fingerprints, blood samples or a lineup appearance, as Schmerber v. California held in 1966. A prosecutor may not comment on a defendant's decision not to testify, Griffin v. California held in 1965. In Doyle v. Ohio in 1976, another Ohio case, the Court held that a defendant's silence after receiving Miranda warnings cannot be used to impeach them at trial. But in Salinas v. Texas in 2013, a plurality held that a person who simply stays silent during voluntary questioning before arrest, without expressly invoking the privilege, can have that silence used against them.

Jesse Morningstar asks whether silence makes a person look guilty. Desjardins says many officers believe it does, but research and experience say otherwise: innocent people stay silent on lawyers' advice all the time, and guilty people often talk at length. An officer who treats silence as proof of guilt is likely to develop tunnel vision, the subject of LES-105's lesson on noble cause corruption. The legal rule and the good investigative habit point the same way: silence is a person's right, and it is not evidence.

Voluntariness under due process

Due process requires that any confession be voluntary. Courts consider the totality of the circumstances: the suspect's age, education, intelligence and mental and physical condition; the length of the interrogation, how it was conducted, and whether the suspect was deprived of food, sleep or medical care; and the use of threats, promises or physical force. Promises of leniency are especially risky. Police deception about the evidence, such as falsely claiming a co-suspect has confessed, is not automatically unconstitutional, as Frazier v. Cupp held in 1969, but it counts in the totality and can push a statement over the line, and several states have recently restricted deception when questioning juveniles. The prosecution must prove voluntariness by a preponderance of the evidence, as Lego v. Twomey held in 1972, although some states require more.

Why innocent people confess

Research into proven wrongful convictions has shown that false confessions are not rare curiosities. In a substantial share of convictions later overturned by DNA evidence, roughly one in four in widely cited counts by innocence organizations, the person had falsely confessed or made incriminating statements. Researchers have identified common patterns. Many false confessions follow long interrogations, often many hours. Many involve young people, people with intellectual disabilities or mental illness, and people who are exhausted, intoxicated or in crisis. Many follow interrogators' claims of overwhelming evidence combined with suggestions that confessing will lead to a better outcome. And many contain details that only the real perpetrator or the police could have known, which turn out to have been fed to the suspect, sometimes unintentionally, during questioning.

The traditional accusatory model of interrogation, associated with the Reid technique developed in the United States in the mid-twentieth century, begins from the assumption that the suspect is guilty and uses confrontation, minimization and maximization to overcome denials. Courts in the United States have criticized aspects of it, and the risks have been documented in many exoneration cases. Since the 1990s many services have moved toward investigative interviewing models such as PEACE, developed in England and Wales, which the class met in LES-104. PEACE treats the interview as a search for information rather than a confession, uses open questions, tests the suspect's account against the evidence, and avoids threats, promises and deception. Desjardins tells the class that the shift was driven as much by results as by law: information-gathering interviews produce more reliable statements and more useful leads.

Undercover officers and statements before formal charges

Miranda does not apply when the suspect does not know they are talking to police. In Illinois v. Perkins in 1990, the Court held that an undercover officer posing as a fellow inmate did not need to give warnings, because the coercive atmosphere Miranda addresses is absent. But once a person has been formally charged, the Sixth Amendment right to counsel attaches, and police may not deliberately elicit statements about the charged crime through informants or undercover officers, as Massiah v. United States held in 1964. That protection is specific to the charged offense; police may still investigate other, uncharged crimes.

Spontaneous statements and recording

Not every statement comes from questioning. In the Lakeshore Hardware case, Kyle Brandt said in the cruiser, without being asked, 'That old lady's tools aren't even worth that much.' Volunteered statements are not the product of interrogation, so Miranda does not apply to them. But if an officer responds with questions designed to draw out more, the exchange becomes interrogation or its functional equivalent, and the Edwards rule applies because Brandt had asked for a lawyer. The officer should record the exact words and the time and should not follow up.

Recording interviews has become standard practice. Most states now require or strongly encourage recording custodial interrogations in serious cases, and federal agencies adopted a presumption of recording in 2014. In Ohio, Revised Code 2933.81 presumes that recorded statements in homicide and serious sexual offense investigations were voluntary, which gives agencies a strong reason to record. Desjardins tells the class that he has never seen a recorded interview hurt an honest officer, and he has seen many unrecorded ones lose cases because no one could prove what happened in the room.

The question-first technique

In Missouri v. Seibert in 2004, police deliberately questioned a suspect without warnings until she confessed, then gave warnings and had her repeat the confession. The Court held that this question-first technique undermines Miranda, and the second statement was inadmissible. By contrast, in Oregon v. Elstad in 1985, where an initial unwarned statement was not the product of a deliberate strategy, a later warned statement was admissible. Officers cannot cure a Miranda violation simply by repeating the questions after warnings.

Checking interview conditions

Port Ardent's major crime unit keeps a log of every custodial interview. Detective Ana Ferreira uses it to check the conditions that courts examine: whether the interview was recorded, how long it lasted and whether breaks were offered. The program flags interviews that need a supervisor's review.

# Port Ardent custodial interviews: (interview id, recorded, minutes, breaks offered)
interviews = [
    ("A-101", True, 75, 1), ("A-102", True, 140, 2), ("A-103", False, 50, 0),
    ("A-104", True, 410, 1), ("A-105", True, 95, 1), ("A-106", True, 260, 3),
    ("A-107", False, 180, 0), ("A-108", True, 35, 0),
]
flags = []
for iid, recorded, minutes, breaks in interviews:
    reasons = []
    if not recorded:
        reasons.append("not recorded")
    if minutes > 240:
        reasons.append("over four hours")
    if minutes > 120 and breaks == 0:
        reasons.append("long with no break")
    if reasons:
        flags.append((iid, reasons))
recorded_count = sum(1 for i in interviews if i[1])
print("recorded:", recorded_count, "of", len(interviews))
for iid, reasons in flags:
    print(iid, "->", ", ".join(reasons))

recorded: 6 of 8
A-103 -> not recorded
A-104 -> over four hours
A-106 -> over four hours
A-107 -> not recorded, long with no break

Six of eight interviews were recorded. Four interviews are flagged: two that were not recorded, one of which also ran three hours with no break, and two that went past four hours, one of them nearly seven. Ferreira reviews each with the investigator. The long interviews are not automatically improper, she says, but they are exactly what defense counsel will examine, so the reasons for their length need to be documented.

Corroborating a confession

Because confessions can be false, good investigators treat them as the beginning of the work, not the end. A reliable confession usually contains details that only the true offender would know and that police had not revealed, sometimes called holdback information. Investigators should decide in advance which details to hold back from the media, witnesses and the suspect, record that decision, and avoid revealing those details during questioning. If the suspect then provides them, the confession gains real weight. If the suspect's account contains only details that were public or that the interviewer mentioned, the confession proves much less.

Desjardins describes the steps his unit followed after any confession in a serious case. They compared the account with the physical evidence, checked every verifiable claim, such as where the suspect said they bought a tool or what time they said they left, and looked for anything in the account that was wrong in a way the true offender would not get wrong. In one case, a man confessed to a break-in but described a back window that the house did not have. The confession was false, given by a frightened man who wanted the questioning to end, and the checks caught it before he was charged.

Jesse asks what happens when a confession and the other evidence conflict. Desjardins says the other evidence should win the investigation's attention: physical evidence, independent witnesses and records do not change under pressure. Officers who find a contradiction must record it and disclose it, because a confession that does not fit the facts is exactly the kind of evidence courts and juries need to weigh with care.