In Confirmation Bias in Criminal Cases, Moa Lidén provides a detailed review of confirmation bias in criminal proceedings. Along with lengthy discussions of the nuances of confirmation bias, Lidén provides striking real-world examples that solidify the reader’s understanding of the concepts discussed. Lidén then suggests strategies for mitigating confirmation bias.
Lidén begins with a review of confirmation bias and why it occurs, explaining that the human brain cannot handle the amount of information that it receives each second, so it takes shortcuts. This is typically an effective and efficient process; however, sometimes these shortcuts lead to errors in judgment (i.e., cognitive biases). She notes that, although the term “confirmation bias” gained acceptance in the 1960s, descriptions of the phenomenon had long existed. Raymond Nickerson provided the most widely accepted definition of confirmation bias in 1998: “the seeking or interpreting of evidence in ways that are partial to existing beliefs, expectations, or a hypothesis in hand” (p 11). Two defining characteristics of confirmation bias are that it is subconscious and the conclusions are premature.
Lidén notes that the field of medicine is not impervious to confirmation bias. Physicians who are aware of a patient’s diagnosis tend to make more diagnosis-consistent interpretations of information than those who are unaware of the diagnosis. Historically, the focus has been on positive results of treatment without considering the possibility that the same patient would have improved without treatment. Confirmation bias is also seen in way that research is published or not published via what she refers to as the “file drawer problem,” the tendency for null findings to remain unpublished. This concept is demonstrated using the example of the physicist Robert Milikan. Milikan won the 1923 Nobel Prize for his work on determining the charge of electrons. He conducted 107 studies, but only 58 of those studies were published, all of which supported his hypothesis. The 59 remaining studies did not support his hypothesis but were not considered a part of his body of work, because they had not been published.
In chapter three, Lidén discusses examples of how confirmation bias affects criminal proceedings. Human decision-making plays a role in nearly every aspect of criminal proceedings. These human decision-makers include police, crime scene investigators, forensic analysts, forensic doctors, prosecutors, and judges. As Lidén notes, “even those who are supposed to be objective can have confirmation bias” (p 3). Further, there is no evidence for judicial exceptionalism, which is the idea that legal actors are somehow more resistant to confirmation bias.
The risk for confirmation bias in criminal proceedings begins with the information-gathering process, including interviews. Confirmation bias influences how evidence is searched for, what evidence is kept, and what evidence is then processed. Interviewers convey subtle clues to witness and defendant interviewees via their body language, providing suggestive information, or asking leading questions. This may occur without the interviewer’s awareness of the behaviors. Witness testimony is widely considered a leading factor in wrongful convictions across the world and is highly susceptible to being guided by confirmation bias.
Forensic analysis can also be influenced by confirmation bias. Knowledge of contextual information, for example, that the suspect has been identified by witnesses, has been found to influence comparisons of types of information that many would otherwise believe not to be prone to error. Examples of such evidence would include shoe prints, bite marks, handwriting samples, blood stain patterns, and more. This influence has been shown in the opinions of forensic physicians, which is particularly problematic, considering the powerful, and at times authoritative, nature of forensic medicine opinions in the courtroom.
Lidén discusses the striking example of the case of Therese Johanessen and Sture Bergwall. Therese, a nine-year-old girl, disappeared in 1988. Her disappearance led to one of the most extensive police inquiries in Norwegian history. These efforts slowly subsided a year later when no suspects had been identified. Then, in 1993, Mr. Bergwall confessed to more than 30 murders, including the murder of Therese. The circumstances should have raised concerns about the accuracy of these confessions, including that the confessions were elicited during therapy, while Mr. Bergwall was in forensic psychiatric care. Further, two individuals he claimed to have killed turned out to be alive. The subsequent investigation focused on obtaining evidence to confirm his guilt. He was convicted of murder and sentenced to continued forensic psychiatric care. Thirteen years later, he retracted his confession. Reexamination of pieces of organic material, initially identified as bone consistent with Therese, concluded that the pieces were not human bone. He was granted a new trial and acquitted.
Examples of confirmation bias often describe an individual who is falsely accused or convicted of a crime. As Lidén notes, confirmation bias can also result in a lack of a conviction when the individual did in fact commit the crime. This occurs because of overconfidence, such that the prosecutor may fail to see and address weaknesses in the case, which would be clear and undeniable to a judge when presented in court. Confirmation bias can change the perception of evidence from being circumstantial and insufficient to being robust and sufficient, which furthers the effects of overconfidence.
Another key concept discussed by Lidén is the potential for a confirmation bias snowball effect. For example, a confession can result in confirmation bias in a forensic analyst, but confirmation bias could also be the reason the suspect confessed in the first place.
Lidén introduces strategies to mitigate confirmation bias. One such strategy is the use of a context manager, who determines what is relevant to share to the individuals involved, such as the forensic analyst. This is termed contextual information management (CIM). Linear sequential unmasking (LSU) is a similar method of minimizing the influence of contextual information, in which contextual information is provided only after an initial evaluation of the data has been made. Further, any deviations in conclusions from those made during the initial assessment must be fully explained and documented. Independent reviewers who are explicitly tasked to review the evidence critically can reduce the potential for confirmation bias. These reviewers are also known as devil’s advocates, red teams, and evidence review boards. Similarly, a second decision-maker who is independent of and blind to the first examiner’s results can be used to reduce confirmation bias. Technological advances and structured evaluations can reduce confirmation bias by reducing the cognitive load.
In summary, confirmation bias exists across the entirety of criminal proceedings because human decision-makers are involved in each step. No individual, regardless of profession, is immune to confirmation bias. It is important for all professionals to understand what confirmation bias is, how it develops, and its potential effects. Strategies such as contextual information management, linear sequential unmasking, independent reviewers, and technology can function to reduce confirmation bias. The concepts discussed in this book are relevant beyond the setting of criminal proceedings. This book would be useful to forensic psychiatrists, general psychiatrists, other mental health professionals, attorneys, judges, and law enforcement personnel seeking to increase their understanding of confirmation bias.
Footnotes
Disclosures of financial or other potential conflicts of interest: None.
- © 2025 American Academy of Psychiatry and the Law
